How Prediction Errors Shape Perception, Attention, and Motivation

By H. D. den Ouden, Peter Kok, F. D. de Lange. 2012

479 citations recorded by the source corpus.

Prediction errors (PE) are a central notion in theoretical models of reinforcement learning, perceptual inference, decision-making and cognition, and prediction error signals have been reported across a wide range of brain regions and experimental paradigms. Here, we will make an attempt to see the forest for the trees and consider the commonalities and differences of reported PE signals in light of recent suggestions that the computation of PE forms a fundamental mode of brain function. We discuss where different types of PE are encoded, how they are generated, and the different functional roles they fulfill. We suggest that while encoding of PE is a common computation across brain regions, the content and function of these error signals can be very different and are determined by the afferent and efferent connections within the neural circuitry in which they arise.

Subfields: Cognitive Neuroscience, Neuroscience, Cognitive Science & Psychology

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

1. Gary Lupyan
2. Andy Clark
3. Alberto Garcia-Garcia
4. Sergiu Oprea
5. Gabriel Kreiman
6. William Lotter
7. David D. Cox
8. James J. DiCarlo

In Neuroscience

1. Gabriel Kreiman
2. William Lotter
3. David D. Cox
4. Gary Lupyan
5. Andy Clark

In Cognitive Science & Psychology

1. David D. Cox
2. Gary Lupyan
3. Andy Clark

Citations in the corpus

• Theory of Mind: A Neural Prediction Problem (2013, neutral): “In IT cortex, predictor neurons code information about object category; error neurons signal mismatches in predicted and observed object category (den Ouden et al., 2012; Peelen and Kastner, 2011).”
• A Review on Deep Learning Techniques for Video Prediction (2020, background): “the mismatch between its predictions and what actually happened [27].”
• Unsupervised Learning of Visual Structure using Predictive Generative Networks (2015, background): “…concerning “predictive coding,” the idea that neuronal systems predict future states of the world and primarily encode only deviations from those predictions (Rao & Ballard, 1999; Summerfield et al., 2006; Den Ouden et al., 2012; Rao & Sejnowski, 2000; Chalasani & Principe, 2013; Zhao et al., 2014).”
• Words and the World (2015, background): “that language input directly affects mental states just like other perceptual inputs (Elman, 2009). For example, in a study by Çukur and colleagues (2013), participants undergoing fMRI watched movie clips passively or while monitoring for humans or for vehicles.”