Using computational theory to constrain statistical models of neural data

doi:10.1101/104737

Paper

Using computational theory to constrain statistical models of neural data

Published Jan 31, 2017 · Scott W. Linderman, S. Gershman

Current Opinion in Neurobiology

Q1 SJR score

27

Citations

1

Influential Citations

Full text Semantic Scholar

Abstract

Computational neuroscience is, to first order, dominated by two approaches: the “bottom-up” approach, which searches for statistical patterns in large-scale neural recordings, and the “top-down” approach, which begins with a theory of computation and considers plausible neural implementations. While this division is not clear-cut, we argue that these approaches should be much more intimately linked. From a Bayesian perspective, computational theories provide constrained prior distributions on neural data—albeit highly sophisticated ones. By connecting theory to observation via a probabilistic model, we provide the link necessary to test, evaluate, and revise our theories in a data-driven and statistically rigorous fashion. This review highlights examples of this theory-driven pipeline for neural data analysis in recent literature and illustrates it with a worked example based on the temporal difference learning model of dopamine.

Study Snapshot

Key takeawayComputational theories can constrain statistical models of neural data, enabling data-driven and statistically rigorous revision of neural models in computational neuroscience.

PopulationOlder adults (50-71 years)

Sample size24

MethodsObservational

OutcomesBody Mass Index projections

ResultsSocial networks mitigate obesity in older groups.

Sign up to use Study Snapshot

Consensus is limited without an account. Create an account or sign in to get more searches and use the Study Snapshot.

Create an account

Paper

Using computational theory to constrain statistical models of neural data

Sign up to use Study Snapshot

References

AIDE: An algorithm for measuring the accuracy of probabilistic inference algorithms

The AIDE algorithm accurately measures the accuracy of approximate inference algorithms, capturing qualitative behavior and detecting failure modes missed by standard heuristics.

Adaptive optimal training of animal behavior

Adaptive optimal training of animal behavior can potentially speed up training processes by accurately inferring learning rules and selecting stimuli that maximize learning rates.

Stan: A Probabilistic Programming Language.

Stan is a probabilistic programming language for specifying statistical models and providing Bayesian inference for continuous-variable models through Markov chain Monte Carlo methods and optimization methods.

Cause for pause before leaping to conclusions about stepping

There is no compelling evidence to support the stepping account of neuron firing rate changes during perceptual decision-making.

Identification of Stable Spike-Timing-Dependent Plasticity from Spiking Activity with Generalized Multilinear Modeling

This study presents a computational framework for quantifying spike-timing-dependent plasticity (STDP) during behavior, which can be applied to behaving animal studies to understand the underlying mechanisms of learning and memory.

Citations

Decoding the application of deep learning in neuroscience: a bibliometric analysis

Deep learning techniques in neuroscience are revolutionizing our understanding of the brain and advancing diagnostics and therapeutic interventions.

Inferring Inference

This study presents a mathematical framework for inferring canonical distributed computations from large-scale neural activity patterns, revealing the brain's internal model of the world and its inference process.

Testing the Tools of Systems Neuroscience on Artificial Neural Networks

Tests on artificial neural networks (ANNs) can help refine systems and cognitive neuroscience tools, potentially accelerating brain research progress.

Functional Connectivity Basis and Underlying Cognitive Mechanisms for Gender Differences in Guilt Aversion

Men exhibit stronger guilt aversion, driven by consideration of social norms, while women show more inequity aversion, with conscientiousness playing a key role in this difference.

How learning unfolds in the brain: toward an optimization view

Understanding inflexible neural variability, multiple learning processes, and large task-nonspecific activity changes is crucial for understanding biological learning in the brain using an optimization framework.

Statistical neuroscience in the single trial limit

Simplifying structure in neural data can reveal novel insights into the trial-by-trial operation of neural circuits in complex and naturalistic animal behaviors.

Parameter Estimation in Multiple Dynamic Synaptic Coupling Model Using Bayesian Point Process State-Space Modeling Framework

Our Bayesian point-process state-space model accurately estimates parameters of dynamic synaptic connections in neuron networks, enabling more accurate characterization of spiking activity.