Traditional one-shot surveys and questionnaires have notable limitations in assessing emotional states. These methods typically provide a single snapshot in time, failing to capture the continuous and dynamic nature of emotions. Additionally, they may not reflect the real-time context in which emotions occur, leading to a loss of valuable information. To address these limitations, more sophisticated approaches, such as mobile app-based digital questionnaires, are necessary. These tools enable ecological momentary assessments, a method for collecting data in real time that allows researchers to continuously track the dynamics and evolution of various states and behaviours. This method provides a more accurate and comprehensive understanding of how emotions fluctuate in natural settings, offering valuable insights for both research and clinical applications. To achieve a longitudinal measurement of people’s emotions, alongside other social and behavioural data, this research developed MobileWell, a digital tool comprising a mobile app and a data analysis platform. The MobileWell app gathers both objective and subjective data from individuals in real time and anonymously. The MobileWell platform automatically and continuously aggregates this data and performs AI-based analyses, providing population-level indicators related to physical activity, social interactions, and emotional states, as well as their links to general well-being.

Despite the intricate dynamic nature of emotional states, there are very few public datasets that measure people’s emotions seamlessly over time. To fill this gap, the MobileWell system was used in this research to collect a novel open dataset. Over one hundred adults were selected to form a representative sample of the Spanish population, balanced in terms of gender, age, location, and annual income. Participants installed the MobileWell app and kept it on their phones for one month, during which they received digital questionnaires several times a day. These questionnaires collected self-reported data related to health status, emotions, and well-being, providing a comprehensive view of the participants’ social, behavioural, emotional, and well-being states. Specifically, this research focused on the assumption that emotional states can be either positive (joy, satisfaction, pleasure) or negative (sadness, frustration, anxiety). Hence, participants were asked six times a day to rate how they felt on a scale from very bad to very good. In terms of psychological well-being, participants were asked to report on several indicators. The Flourishing Scale was used to measure self-perceived success, while the Brief Resilience Scale captured the ability to recover from stress. The Satisfaction with Life Scale assessed perceived life satisfaction. The Generalized Anxiety Disorder Scale  measured anxiety-related symptoms, and the Patient Health Questionnaire-9 evaluated the frequency of various prominent depression symptoms. Finally, the Acceptance and Action Questionnaire-II measured the level of experiential avoidance. This comprehensive approach provides a detailed understanding of the participants’ emotional and psychological well-being.

Certain influential theories of emotion propose that individuals have a characteristic baseline emotional state that serves as a reference point to which they return after intense emotional experiences or internal regulatory events, such as the ones experienced during major life transitions or stressful situations. This research explored the alternative hypothesis that people have more than one characteristic emotional state and experience abrupt transitions between these states. This hypothesis was inspired by the concept of multistability observed in other complex natural systems, such as lakes, which can undergo sudden shifts between clear and turbid water states, or tropical forests, that can rapidly transform into savannas. To investigate this hypothesis, this research analysed the emotional data collected in the MobileWell100+ study. A stability landscape analysis approach, originally developed for ecological systems, was applied to the emotional time series data. The results of the analysis revealed that more than half of the participants (61.5%) exhibited bistable behaviour, meaning their emotions often switched abruptly between positive and negative states, rather than returning to a stable baseline. Moreover, this emotional bistability is observed almost equally in women (47.5%) and men (52.5%). Understanding that people are generally bistable has significant implications for psychological research and clinical practice. It underscores the importance of considering the complexity and variability of emotional experiences when developing interventions and supports for mental health. In addition, recognising bistable emotional patterns could lead to more effective strategies for promoting psychological well-being and resilience.

This research further examined whether transitions between emotional states are associated with people’s well-being. The first step was to quantify how intensely participants shifted between positive and negative emotional states. Using this data, several novel metrics were developed to measure these shifts, finding that a particular metric that quantifies the propensity to shift from a positive to a negative emotional state (the positive-to-negative affect shift ratio, P2N-ASR) was a strong predictor of various psychological well-being indicators, including depression, anxiety, life satisfaction, and resilience. The P2N-ASR outperformed traditional metrics like mean and standard deviation levels of positive and negative affect in predicting some of these indicators. Higher levels of P2N-ASR were particularly associated with worse well-being, such as increased depression and anxiety symptoms and reduced resilience and life satisfaction. Unlike previous methods that require complex emotional ratings, the P2N-ASR relies on a simple question about whether individuals feel good or bad. This simplicity, combined with its strong predictive value, makes the P2N-ASR a useful tool for mental health practitioners. The authors’ discovery of bistability in human emotions, a phenomenon they termed “affective bistability”, is crucial for a more accurate understanding of both healthy and disordered emotional experiences.

This research uncovers an intricate view of emotional dynamics, revealing that a majority of the sample exhibited bistable behaviour, with their emotions abruptly shifting between positive and negative states. This challenges the traditional notion of emotional stability, which assumes emotions generally return to a baseline after disturbances. Using the MobileWell app and a novel metric known as the positive-to-negative affect shift ratio (P2N-ASR), the research highlights that the propensity for such emotional shifts is a strong predictor of challenges to psychological well-being. Specifically, higher P2N-ASR levels are linked to increased anxiety and depression and lower levels of resilience and life satisfaction. The simplicity of the P2N-ASR—based on a straightforward assessment of whether individuals feel good or bad—makes it a highly valuable tool for mental health professionals. It outperforms traditional metrics that rely on complex ratings of emotional states, offering a more practical and insightful approach to tracking and predicting mental health. This simple classification of good or bad aligns with how people naturally communicate their affective state in everyday interactions, making it more accessible and easier to convey, an aspect that is of great value in clinical practice. The discovery of “affective bistability” provides a deeper understanding of emotional experiences, suggesting that mental health interventions could benefit from addressing the dynamic nature of emotions rather than focusing solely on baseline stability.

BAÑOS, O.; BAILON, C.; DAMAS, M.; GOICOECHEA, C.; PERAKAKIS, P.; POMARES, H.; RODRIGUEZ-LEON, C.; SANABRIA, D.; VILLALONGA, C. (2022): MobileWell100+: A Multivariate Longitudinal Mobile Dataset for Investigating Individual and Collective Well-Being [dataset]. Zenodo. 

BAÑOS, O., et al. (2021): MobileWell (versión 1.0) [software]. 

DAKOS, V.; KÉFI, S. (2022): «Ecological resilience: what to measure and how». En: Environmental Research Letters, 17(4), 043003.

HIROTA, M.; HOLMGREN, M.; VAN NES, E. H.; SCHEFFER, M. (2011): «Global resilience of tropical forest and savanna to critical transitions». En: Science, 334(6053), 232-235.

STINSON, L.; LIU, Y.; DALLERY, J. (2022): «Ecological momentary assessment: a systematic review of validity research». En: Perspectives on Behavior Science, 45(2), 469-493.