152 research outputs found
A Bayesian foundation for individual learning under uncertainty
No full textComputational learning models are critical for understanding mechanisms of adaptive behavior. However, the two major current frameworks, reinforcement learning (RL) and Bayesian learning, both have certain limitations. For example, many Bayesian models are agnostic of inter-individual variability and involve complicated integrals, making online learning difficult. Here, we introduce a generic hierarchical Bayesian framework for individual learning under multiple forms of uncertainty (e.g., environmental volatility and perceptual uncertainty). The model assumes Gaussian random walks of states at all but the first level, with the step size determined by the next higher level. The coupling between levels is controlled by parameters that shape the influence of uncertainty on learning in a subject-specific fashion. Using variational Bayes under a mean field approximation and a novel approximation to the posterior energy function, we derive trial-by-trial update equations which (i) are analytical and extremely efficient, enabling real-time learning, (ii) have a natural interpretation in terms of RL, and (iii) contain parameters representing processes which play a key role in current theories of learning, e.g., precision-weighting of prediction error. These parameters allow for the expression of individual differences in learning and may relate to specific neuromodulatory mechanisms in the brain. Our model is very general: it can deal with both discrete and continuous states and equally accounts for deterministic and probabilistic relations between environmental events and perceptual states (i.e., situations with and without perceptual uncertainty). These properties are illustrated by simulations and analyses of empirical time series. Overall, our framework provides a novel foundation for understanding normal and pathological learning that contextualizes RL within a generic Bayesian scheme and thus connects it to principles of optimality from probability theory
Editorial note to: Hierarchical prediction errors in midbrain and basal forebrain during sensory learning
No full textIn this issue, we, the editors of Neuron, are publishing the Correction from Sandra Iglesias, Klaas Enno Stephan, and colleagues, which explains how the error in their Neuron manuscript (Iglesias et al., 2013, Neuron 80, 519-530) arose and the effects of the error. Our decision to publish the Correction is based on the nature of the error, which was one of computational analysis. The authors failed to switch off the default code in their main analytical software, and this error subsequently impacted all downstream analysis and thus multiple figures and tables. After discussion of the error and the changes after application of the correct analysis amongst the Neuron editorial team, the Cell Press Editorial Department leadership team, and external experts in the field, we agreed that this type of mistake is not uncommon in fMRI analysis, and as in this case the main conclusions of the study remain intact, it would serve the community to publish a detailed and transparent correction that explains the error and presents the corrected figures using the intended analysis that was originally reported in the paper. We respect and appreciate the authors’ step in coming forward to bring the error to the attention of the field
Analysis of anatomical and effective connectivity in neural systems
No full textEThOS - Electronic Theses Online ServiceGBUnited Kingdo
Untersuchungen zur funktionellen Konnektivität des Gehirns
No full textThis dissertation includes two independent studies that investigate two complementary aspects of functional connectivity, i .e. context-invariance and context-specificity, in the Macaque and the human brain. In the first study, a computational meta-analysis of published electrophysiological data an context-independent functional brain connectivity was conducted by means of three independent methods. Almost 4,000 individual experimental findings an interactions between areas of the Macaque cortex (obtained by strychnine neuronography) were analyzed. The independent analyses gave compatible results and showed that (i) the network of functional interactions between cortical areas showed clear small world" characteristics (i.e. a strongly clustered structure with short average path length), and that (ii) this structure contained three main functional clusters, i.e. sensorimotor, visual, and orbito-temporo-insular groups of areas. This study thus provided first evidence for a functional small world" architecture of the primate cortical network. This type of architecture had previously been postulated in the literature, but had not been directly demonstrated. The second study investigated the effects of the atypical antipsychotic substance olanzapine an the functional connectivity of the cerebellum during a simple motor task (self-paced finger tapping). Six schizophrenic patients and six control subjects matched for age and sex were investigated by functional magnetic resonance imaging (fMRI) twice. At the time of the first scan, patents were not medicated; the second scan took place after three weeks of medication by olanzapine. The analyses of the fMRI data showed that, in the context of the investigated motor task, (i) olanzapine led to pronounced changes of cerebellar functional connectivity (CFC) in various brain regions. A major part of three changes were found throughout the prefrontal cortex and the mediodorsal thalamus. These findings are relevant for the concept of "cognitive dysmetria". (ii) significant CFC changes in motor areas were found both within the patient group after medication as well as in the non-medicated control group after repetition of the experiment. Therefore, they may correspond to unspecific repetition effects rather than effects due to olanzapine. In a subsequent analysis that took the repetition into account no significant CFC changes were found in motor regions. (iii) olanzapine normalized the CFC patterns of the patients for the right, but not for the left cerebellum. This study provided the first experimental data an the effects of atypical antipsychotic agents an functional brain connectivity and demonstrated pronounced olanzapine-dependent changes of functional couplings between cerebellum, thalamus, and prefrontal cortex
Computational approaches to psychiatry
No full textA major reason for disappointing progress of psychiatric diagnostics and nosology is the lack of tests which enable mechanistic inference on disease processes within individual patients. The resulting inability to pursue formal differential diagnosis has forced the field to stick to symptom-based diagnostic schemes with limited predictive validity concerning treatment response and clinical outcome. A promising new approach is the use of computational modeling for inferring mechanisms which generate observed behavior and brain activity in psychiatric patients. However, while this computational approach to psychiatry is rapidly gaining attention, much work remains to be done to finesse existing computational models, making them 'fit for practice' in a clinical setting and proving their validity in longitudinal studies. This review outlines recent methodological advances and strategies in this regard, focusing on generative models which infer mechanistically interpretable parameters (of computational or physiological processes) from measured behavior and brain activity. © 2013 Elsevier Ltd
The history of CoCoMac
No full textAbstractCoCoMac, the “Collation of Connectivity Data for the Macaque” is a relational database system which presently constitutes the largest electronic repository of published neuroanatomical connectivity data. Developed since 1996, CoCoMac comprises approximately 40,000 experimental findings on anatomical connections in the macaque brain, as derived from neuroanatomical tract tracing studies. In this historical review, I describe the origin and the history of CoCoMac from a personal perspective, illustrate the principles of its structure and outline the impact it has had on systems neuroscience, in particular as a prelude to the “Human Connectome” research programme
Uncertainty in perception and the Hierarchical Gaussian Filter.
No full textIn its full sense, perception rests on an agent's model of how its sensory input comes about and the inferences it draws based on this model. These inferences are necessarily uncertain. Here, we illustrate how the Hierarchical Gaussian Filter (HGF) offers a principled and generic way to deal with the several forms that uncertainty in perception takes. The HGF is a recent derivation of one-step update equations from Bayesian principles that rests on a hierarchical generative model of the environment and its (in)stability. It is computationally highly efficient, allows for online estimates of hidden states, and has found numerous applications to experimental data from human subjects. In this paper, we generalize previous descriptions of the HGF and its account of perceptual uncertainty. First, we explicitly formulate the extension of the HGF's hierarchy to any number of levels; second, we discuss how various forms of uncertainty are accommodated by the minimization of variational free energy as encoded in the update equations; third, we combine the HGF with decision models and demonstrate the inversion of this combination; finally, we report a simulation study that compared four optimization methods for inverting the HGF/decision model combination at different noise levels. These four methods (Nelder-Mead simplex algorithm, Gaussian process-based global optimization, variational Bayes and Markov chain Monte Carlo sampling) all performed well even under considerable noise, with variational Bayes offering the best combination of efficiency and informativeness of inference. Our results demonstrate that the HGF provides a principled, flexible, and efficient-but at the same time intuitive-framework for the resolution of perceptual uncertainty in behaving agents
Model selection and gobbledygook: Response to Lohmann et al
No full textLohmann et al. (in the same issue) make three unremarkable observations about model selection and use them to critique dynamic causal modelling—a Bayesian model selection procedure based on causal models of dynamical systems (Marreiros et al., 2010). In this response, we unpack their misconceptions and try to answer their questions
- …
