Probabilistic brains: knowns and unknowns

By Alexandre Pouget, Jeffrey M. Beck, Wei Ji Ma, Peter E. Latham. Nature Neuroscience, 2013

641 citations recorded by the source corpus.

There is strong behavioral and physiological evidence that the brain both represents probability distributions and performs probabilistic inference. Computational neuroscientists have started to shed light on how these probabilistic representations and computations might be implemented in neural circuits. One particularly appealing aspect of these theories is their generality: they can be used to model a wide range of tasks, from sensory processing to high-level cognition. To date, however, these theories have only been applied to very simple tasks. Here we discuss the challenges that will emerge as researchers start focusing their efforts on real-life computations, with a focus on probabilistic learning, structural learning and approximate inference.

Subfields: Cognitive Neuroscience

View Paper: https://www.semanticscholar.org/paper/bbab6293783b741729f4f1636b9267e085aa329f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

Citations in the corpus

Correlations and Neuronal Population Information (2016, background): “Different proposals exist for how such probabilistic representations can be achieved, and these may create different patterns of correlations (see Pouget et al. 2013 for a review).”
The free energy principle for action and perception: A mathematical review (2017, background): “This assumption implies that an organism only represents the expectation value of environmental variables, and not a distribution over states, see [38, 49, 50] for a 815 nice descriptions of this assumption.”
Amortized Inference in Probabilistic Reasoning (2014, background): “One view of probabilistic reasoning holds that our brains are equipped with general-purpose inference algorithms that can be used to answer arbitrary queries (Griffiths et al., 2012; Pouget et al., 2013).”
Distributional reinforcement learning in the brain (2020, background): “For example, probabilistic population codes (PPCs) [73,74] and distributed distributional codes (DDCs) [75,76] employ population coding schemes from which various statistical parameters of a distribution can be read out, making them parametric codes .”
Robots that can adapt like animals (2014, methodology): “…optimization procedure followed by our robot appears similar to the technique employed by humans when they optimize an unknown function 23 , and there is strong evidence that animal brains learn probability distributions, combine them with prior knowledge, and act as Bayesian optimizers 25,26 .”
Cognitive computational neuroscience (2018, background): “The representation of probability distributions in neuronal populations has been explored theoretically and experimentally.(161,162) However, relating Bayesian inference and learning, especially structure learning in nonparametric models, to its implementation in the brain remains challenging.”
Noise as a Resource for Computation and Learning in Networks of Spiking Neurons (2014, background): “The more frequently proposed hypothesis is that such distributions are encoded there in an arithmetical form, and that knowledge is extracted from such probabilistic knowledge through arithmetical operations such as belief propagation [37], [38].”
The Hamiltonian Brain: Efficient Probabilistic Inference with Excitatory-Inhibitory Neural Circuit Dynamics (2014, background): “The apparent success of probabilistic inference in accounting for a diverse set of experimental observations raises the question of how neural systems might represent and compute with uncertainty [17].”
Deep neural networks: a new framework for modelling biological vision and brain information processing (2015, background): “The brain clearly handles uncertainty well in many contexts (Tenenbaum et al. 2006; Pouget et al. 2013), so it is helpful to view its inferences as approximations, however rough, to rational probabilistic inference.”