Bayesian Computation Framework Research Articles

Sticking to the Evidence? A Behavioral and Computational Case Study of Micro-Theory Change in the Domain of Magnetism.

Constructing an intuitive theory from data confronts learners with a "chicken-and-egg" problem: The laws can only be expressed in terms of the theory's core concepts, but these concepts are only meaningful in terms of the role they play in the theory's laws; how can a learner discover appropriate concepts and laws simultaneously, knowing neither to begin with? We explore how children can solve this chicken-and-egg problem in the domain of magnetism, drawing on perspectives from computational modeling and behavioral experiments. We present 4- and 5-year-olds with two different simplified magnet-learning tasks. Children appropriately constrain their beliefs to two hypotheses following ambiguous but informative evidence. Following a critical intervention, they learn the correct theory. In the second study, children infer the correct number of categories given no information about the possible causal laws. Children's hypotheses in these tasks are explained as rational inferences within a Bayesian computational framework.

Quantitative reconstruction of seasonality from stable isotopes in teeth

Seasonal variation in rainfall and temperature are major determinants of ecosystem structure and productivity and influence physical processes such as erosion rates and glaciation. By extension, seasonality can drive evolutionary change, and is hypothesized to have influenced early human subsistence and technological development. Rainfall patterns are reflected by oxygen isotope ratios (δ18O values) in mammalian tooth enamel, which records environmental chemistry as reflected by blood chemistry during mineralization. Fossilized herbivore molars are commonly used to study paleoseasonality, since temporal variation in precipitation δ18O results in spatial variation in tooth mineral δ18O. However, this approach has been hampered by incomplete knowledge of isotope incorporation during tooth mineralization. Here we test a new synchrotron-based mineralization model and demonstrate its potential to quantitatively reconstruct original seasonal input histories under a Bayesian computational framework. To accomplish this, we first integrate the mineralization model with blood water oxygen isotope turnover to produce high-resolution spatial δ18O predictions. We test these predictions using fine-scaled tooth enamel phosphate δ18O measurements (n = 109 locations sampled in a 2D grid format) in a sheep (Ovis aries) subjected to a controlled water switch (switch δ18O magnitude of 12‰). Tooth mineralization, blood oxygen turnover and isotopic measurements demonstrate that enamel secretion and maturation waves advance at nonlinear rates with distinct geometries. Final enamel isotopic composition may be influenced by isotopic shifts during enamel maturation from amorphous precursors. We combine these observations to produce a Bayesian inverse model for reconstructing water δ18O inputs from tooth isotopic measurements. The system accurately reproduces the controlled switch of the experimental animal and reveals transient documented meteorological events that were unplanned during the experiment. A simulation approach shows that given appropriate priors and search methodologies, this system also reconstructs precipitation histories at different latitudes with striking fidelity. We show that in some contexts even simple sampling approaches (i.e. 1D transects) can yield accurate reconstructions of seasonality. Finally, we refine our mineralization model further by exploring the possible extent of isotopic resetting during mineral phase transitions. This work demonstrates that tooth isotopic measurements can be employed to reconstruct past seasonal isotopic variation in rainfall quantitatively, allowing exploration of the relationships among climate, earth history and evolution.

A Stochastic Model for Assessing Chlamydia trachomatis Transmission Risk Using Longitudinal Observational Data.

Bacterium Chlamydia trachomatis causes genital chlamydia infection. Yet little is known about the transmission efficiency of this organism. Ethical constraint against exposing healthy subjects to infected partners precludes the possibility of quantifying transmission risk through controlled experiments. This research proposes an alternative strategy that relies on observational data. Specifically, we present a stochastic model that treats longitudinally observed infection states in a group of young women as a Markov process. The proposed model explicitly accommodates the parameters of C. trachomatis transmission, including per-encounter sexually transmitted infection (STI) acquisition risks, with and without condom protection, and the probability of antibiotic treatment failure. The male-to-female transmission probability of C. trachomatis is then estimated by combining the per-encounter disease acquisition risk and the organism's prevalence in the male partner population. The proposed model is fitted in a Bayesian computational framework.