The dimensionality of neural representations for control

By David Badre, Apoorva Bhandari, Haley Keglovits, Atsushi Kikumoto. 2020

134 citations recorded by the source corpus.

Cognitive control allows us to think and behave flexibly based on our context and goals. At the heart of theories of cognitive control is a control representation that enables the same input to produce different outputs contingent on contextual factors. In this review, we focus on an important property of the control representation’s neural code: its representational dimensionality. Dimensionality of a neural representation balances a basic separability/generalizability trade-off in neural computation. We will discuss the implications of this trade-off for cognitive control. We will then briefly review current neuroscience findings regarding the dimensionality of control representations in the brain, particularly the prefrontal cortex. We conclude by highlighting open questions and crucial directions for future research.

Subfields: Cognitive Neuroscience

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

Orthogonal representations for robust context-dependent task performance in brains and neural networks (2022, neutral): “However, how the neural representations that support context-dependent task performance are acquired remains a key open question for cognitive and neural scientists (Gao and Ganguli, 2015; Mastrogiuseppe andOstojic, 2018; Cueva et al., 2020; Dubreuil et al., 2020; Badre et al., 2021; Freund et al., ”
Interpreting neural computations by examining intrinsic and embedding dimensionality of neural activity (2021, background): “, the ability to flexibly use the underlying representation in the context of different policies, plans and action [71,72].”
Neural Coding of Cognitive Control: The Representational Similarity Analysis Approach. (2020, background): “, abstract rules [3,40], high-dimensional [41] task conjunctions [42,43], compositional subtasks [44,45], attentional templates [46], hierarchical task schemas [47–49]).”
Hierarchical Reconfiguration of Neurocognitive Task Set Representations Mediates Cognitive Flexibility (2026, unknown): “Understanding how these demands are jointly supported requires examining the representational structures encoded in the brain that implement cognitive control (Monsell, 2003; Meiran et al., 2000; Rogers & Monsell, 1995; Badre et al., 2021).”
Novel Verbal Instructions Recruit Abstract Neural Patterns of Time-Variable Information Dimensionality (2025, background): “A highly relevant descriptor of control representations is their underlying neural geometry (Badre et al., 2021) and their balance between neural flexibility and stability (Bernardi et al., 2020; Bhandari et al., 2024).”
Radical flexibility of neural representation in frontoparietal cortex and the challenge of linking it to behaviour (2024, unknown): “Given the inherent cost of task switching and the trade-off between generalisability and flexibility of representation [57,58], a radically flexible system may also avoid strong prioritisation in neural coding when task demands are low [33,42].”
Aperiodic and oscillatory systems underpinning human domain-general cognition (2024, unknown): “Please explain this point.”
Cognition is an emergent property (2024, unknown): “Instead, population spiking activity is restricted to a low-dimensional subspace of the full high-dimensional population space (Figure 3c) [37–40].”
Distinct functions for beta and alpha bursts in gating of human working memory (2023, background): “This shapes low-dimensional and task-related aspects of single neuronal spiking that are needed to solve the cognitive task at hand 40,41 .”