Canonical neural networks perform active inference

This work considers a class of canonical neural networks comprising rate coding models, wherein neural activity and plasticity minimise a common cost function—and plasticity is modulated with a certain delay. We show that such neural networks implicitly perform active inference and learning to minimise the risk associated with future outcomes. Mathematical analyses demonstrate that this biological optimisation can be cast as maximisation of model evidence, or equivalently minimisation of variational free energy, under the well-known form of a partially observed Markov decision process model. This equivalence indicates that the delayed modulation of Hebbian plasticity—accompanied with adaptation of firing thresholds—is a sufficient neuronal substrate to attain Bayes optimal inference and control. We corroborated this proposition using numerical analyses of maze tasks. This theory offers a universal characterisation of canonical neural networks in terms of Bayesian belief updating and provides insight into the neuronal mechanisms underlying planning and adaptive behavioural control.

Low-frequency atmospheric variability patterns and synoptic types linked to large floods in the lower Ebro River basin

This study analyzes the atmospheric variability that caused the largest floods affecting the town of Tortosa in the mouth of the Ebro River (northeast Iberian Peninsula). The Tortosa flood database and flood marks in the nearby town of Xerta are used to define the more relevant flooding episodes (discharges > 2900 m3s−1) of the 1600-2005 period. We explore the atmospheric variability based on low-frequency patterns and synoptic types applying a multivariable analysis to grids at sea-level pressure and geopotential at 500 hPa provided by the 20th Century V3 Reanalysis Project for the instrumental period (since 1836). Output from the Last Millennium Ensemble Project was used to analyze the sea-level pressure over the pre-instrumental period (before 1836). Our analysis includes 33 flood episodes. Four synoptic types are related to floods in Tortosa since 1836, characterized by low-pressure systems that interact with the Mediterranean warm air-mass and promote the atmosphere destabilization. Flooding in Tortosa is related to relative high values of solar activity, positive Northern Hemisphere temperature anomalies and NAO in positive phase. This indicates that the major floods are related to zonal atmospheric circulations (west to east cyclone transfer). During winter, the main impact of the floods is located at the western part of the basin, while the Pyrenean sub-basins are affected during autumn. The major finding is that similar flood behavior is detected since 1600, improving our understanding of past climates, enhancing the knowledge base for some aspects and impacts of climate change and reducing uncertainty about future outcomes.

The Predictive Utility of Valuing the Future for Smoking Cessation: Findings from the ITC 4 Country Surveys

Background: Delay discounting (DD) and time perspective (TP) are conceptually related constructs that are theorized as important determinants of the pursuit of future outcomes over present inclinations. This study explores their predictive relationships for smoking cessation. Methods: 5006 daily smokers at a baseline wave provided 6710 paired observations of quitting activity between two waves. Data are from the International Tobacco Control (ITC) smoking and vaping surveys with samples from the USA, Canada, England, and Australia, across three waves conducted in 2016, 2018 and 2020. Smokers were assessed for TP and DD, plus smoking-specific predictors at one wave of cessation outcomes defined as either making a quit attempt and/or success among those who tried to quit which was ascertained at the subsequent survey wave. Results: TP and DD were essentially uncorrelated. TP predicted making quit attempts, both on its own and controlling for other potential predictors but was negatively associated with quit success. By contrast, DD was not related to making quit attempts, but high DD predicted relapse. The presence of financial stress at baseline resulted in some moderation of effects. Conclusions: Understanding the mechanisms of action of TP and DD can advance our understanding of, and ability to enhance, goal-directed behavioural change. TP appears to contribute to future intention formation, but not necessarily practical thought of how to achieve goals. DD is more likely an index of capacity to effectively generate competing future possibilities in response to immediate gratification.

Trajectory Curves for Purposes of Benchmarking and Predicting Clinical Outcomes: A Scoping Review

Trajectory curves are valuable tools to benchmark patient health status and predict future outcomes. A longitudinal study is underway to examine social participation after burn injury using the Life Impact Burn Recovery Evaluation (LIBRE) Profile with the goal of developing trajectory curves for specific domains that focus on social re-integration. We conducted a scoping review to inform and understand trajectory curves applied in clinical settings to compare outcomes for an individual to a matched cohort of comparable patients or predicted expected outcomes over time. This scoping review utilized a PubMed search from January 2014 to August 2019 for the following terms: “trajectory curves” or “trajectory models” and “clinic” or “clinical.” Only articles that specifically referenced longitudinal and clinical research designs were included in the scoping review. Articles were assessed using standard scoping review methods and categorized based on clinical application of trajectory curves for either benchmarking or prediction. The initial literature review identified 141 manuscripts and 34 met initial inclusion criteria. The reviewed articles support the clinical use of trajectory curves. Findings provide insight into several key determinants involved with the successful development and implementation of trajectory curves in clinical settings. These findings will inform efforts to use the LIBRE Profile to model social participation recovery and assist in developing effective strategies using trajectory curves to promote social reintegration after burn injury.

Sociality of Future Outcomes Moderates The Effects Of Warmth And Competence On Social Optimism Bias - A Pre-Registered Study

People are overoptimistic about the future of those they like or admire (social optimism bias), expecting significantly more desirable than undesirable outcomes. By contrast, they are pessimistic about those they don’t like. To operationalize the (dis)like of social targets, warmth and competence are used as two universal dimensions of social perception. In this pre-registered study, we replicate previous findings while adding two new levels of complexity. First, we make the distinction between the sociality of future outcomes: “alone” outcomes (e.g., enjoying a quiet afternoon by oneself) and “social” outcomes (e.g., enjoying a vacation with the significant other). Second, we investigate the effect of attachment styles on one’s expectations for alone and social outcomes towards the social targets. In line with our hypotheses, the sociality of outcomes moderates both the additive and the multiplicative effects of the warmth and competence of social targets on social optimism bias. Diverging from our hypotheses, we find that attachment anxiety and avoidance do not influence the effects of warmth and competence on social optimism bias. However, exploratory analyses revealed that attachment dimensions buffer the magnitude of social optimism bias for highly self-relevant social targets but do not impact a social pessimism bias for irrelevant targets.

Forecasting Facing Economic Shifts, Climate Change and Evolving Pandemics

By its emissions of greenhouse gases, economic activity is the source of climate change which affects pandemics that in turn can impact badly on economies. Across the three highly interacting disciplines in our title, time-series observations are measured at vastly different data frequencies: very low frequency at 1000-year intervals for paleoclimate, through annual, monthly to intra-daily for current climate; weekly and daily for pandemic data; annual, quarterly and monthly for economic data, and seconds or nano-seconds in finance. Nevertheless, there are important commonalities to economic, climate and pandemic time series. First, time series in all three disciplines are subject to non-stationarities from evolving stochastic trends and sudden distributional shifts, as well as data revisions and changes to data measurement systems. Next, all three have imperfect and incomplete knowledge of their data generating processes from changing human behaviour, so must search for reasonable empirical modeling approximations. Finally, all three need forecasts of likely future outcomes to plan and adapt as events unfold, albeit again over very different horizons. We consider how these features shape the formulation and selection of forecasting models to tackle their common data features yet distinct problems.

The Prognostic Utility of Personality Traits Versus Past Psychiatric Diagnoses: Predicting Future Mental Health and Functioning

Past psychiatric diagnoses are central to patient case formulation and prognosis. Recently, alternative classification models such as the Hierarchical Taxonomy of Psychopathology (HiTOP) proposed to assess traits to predict clinically relevant outcomes. In the current study, we directly compared personality traits and past diagnoses as predictors of future mental health and functioning in three independent, prospective samples. Regression analyses found that personality traits significantly predicted future first onsets of psychiatric disorders (change in [∆] R2 = .06–.15), symptom chronicity (∆ R2 = .03–.06), and functioning (∆ R2 = .02–.07), beyond past and current psychiatric diagnoses. Conversely, past psychiatric diagnoses did not provide an incremental prediction of outcomes when personality traits and other concurrent predictors were already included in the model. Overall, personality traits predicted a variety of outcomes in diverse settings beyond diagnoses. Past diagnoses were generally not informative about future outcomes when personality was considered. Together, these findings support the added value of personality traits assessment in case formulation, consistent with the HiTOP model.

Schizotypy, Lifestyle Behaviors, and Health Indicators in a Young Adult Sample

Problematic lifestyle behaviors and high rates of physical illness are well documented in people with schizophrenia, contributing to premature mortality. Yet, there is a notable absence of research examining general lifestyle and health issues in participants at risk for psychosis. This form of research may help identify concerns that exist during prodromal periods related to future outcomes. Accordingly, the current study examined lifestyle and health in a nonclinical sample of 530 young adults with varying levels of schizotypy. Increasing symptom severity was associated with greater somatic symptoms and poorer sleep quality across positive, negative, and disorganized domains. Elevated negative and disorganized symptoms were associated with significantly reduced health-related quality of life, while evidence for reduced engagement in health behaviors was largely limited to those with elevated negative schizotypy. No relationships emerged between symptom presentation/severity and body mass index or substance use, although zero-order correlations suggested an association between disorganized schizotypy and nicotine use. The pattern of relationships in the current study was consistent with findings from the ultra-high risk and clinical literature suggesting that lifestyle and health concerns may exist on a continuum with psychosis. Future research should seek to clarify if these patterns are associated with long-term physical or mental health outcomes.

Constructing Epidemiologic Cohorts from Electronic Health Record Data

In the United States, electronic health records (EHR) are increasingly being incorporated into healthcare organizations to document patient health and services rendered. EHRs serve as a vast repository of demographic, diagnostic, procedural, therapeutic, and laboratory test data generated during the routine provision of health care. The appeal of using EHR data for epidemiologic research is clear: EHRs generate large datasets on real-world patient populations in an easily retrievable form permitting the cost-efficient execution of epidemiologic studies on a wide array of topics. Constructing epidemiologic cohorts from EHR data involves as a defining feature the development of data machinery, which transforms raw EHR data into an epidemiologic dataset from which appropriate inference can be drawn. Though data machinery includes many features, the current report focuses on three aspects of machinery development of high salience to EHR-based epidemiology: (1) selecting study participants; (2) defining “baseline” and assembly of baseline characteristics; and (3) follow-up for future outcomes. For each, the defining features and unique challenges with respect to EHR-based epidemiology are discussed. An ongoing example illustrates key points. EHR-based epidemiology will become more prominent as EHR data sources continue to proliferate. Epidemiologists must continue to improve the methods of EHR-based epidemiology given the relevance of EHRs in today’s healthcare ecosystem.

Which Methods Are Useful to Justify Public Policies? An Analysis of Cost–Benefit Analysis, Multi-Criteria Decision Analysis, and Non-Aggregate Indicator Systems

Science-based methods for assessing the practical rationality of a proposed public policy typically represent assumed future outcomes of policies and values attributed to these outcomes in an idealized, that is, intentionally distorted way and abstracted from aspects that are deemed irrelevant. Different types of methods do so in different ways. As a consequence, they instantiate the properties that result from abstraction and idealization such as conceptual simplicity versus complexity, or comprehensiveness versus selectivity of the values under consideration to different degrees. I hold that none of these methods is best in general. Instead, I opt for the valuation method that is useful for the policy issue in question both in terms of its relevance and in terms of its practicability. Relevance requires that the method can represent and account for what is at stake in the policy issue. Practicability refers to aspects such as easy versus difficult handling of the method. To argue for the claim, I evaluate three types of valuation methods: (1) cost–benefit analysis that rests on unidimensional measurement and ranking, (2) multi-criteria decision analysis that applies multi-dimensional measurement but unidimensional ranking, and (3) non-aggregate indicator systems that operate with multi-dimensional measurement and sometimes also multi-dimensional ranking. Second-order justification indicating whether and how the valuation method chosen is capable of accounting for the substantive value considerations that constitute the real-world policy issue in question renders the conditions on which the results of a proposed policy evaluation rest transparent.