Inferences on a multidimensional social hierarchy use a grid-like code

Park, Seongmin A.; Miller, Douglas S.; Boorman, Erie D.

doi:10.1038/s41593-021-00916-3

Cited by 131 publications

(115 citation statements)

References 76 publications

Supporting

Mentioning

105

Contrasting

Order By: Relevance

“…This idea fits well with a general framework where the cost of switching must be traded off against the strength of control: stronger control results in less catastrophic forgetting, but more difficulty switching (Herd, Banich, & O'Reilly, 2006;Shenhav, Botvinick, & Cohen, 2013). We perform our simulations on a task designed to induce learning of map-like representations (Park, Miller, & Boorman, 2021;Park, Miller, Nili, Ranganath, & Boorman, 2020;Russin, Zolfaghar, Park, Boorman, & O'Reilly, 2021) so that we could additionally investigate how cognitive control affected the model's representations.…”

Section: Introductionmentioning

confidence: 81%

A Neural Network Model of Continual Learning with Cognitive Control

Russin¹,

Zolfaghar²,

Park³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Neural networks struggle in continual learning settings from catastrophic forgetting: when trials are blocked, new learning can overwrite the learning from previous blocks. Humans learn effectively in these settings, in some cases even showing an advantage of blocking, suggesting the brain contains mechanisms to overcome this problem. Here, we build on previous work and show that neural networks equipped with a mechanism for cognitive control do not exhibit catastrophic forgetting when trials are blocked. We further show an advantage of blocking over interleaving when there is a bias for active maintenance in the control signal, implying a tradeoff between maintenance and the strength of control. Analyses of map-like representations learned by the networks provided additional insights into these mechanisms. Our work highlights the potential of cognitive control to aid continual learning in neural networks, and offers an explanation for the advantage of blocking that has been observed in humans.

show abstract

Section: Introductionmentioning

confidence: 81%

A Neural Network Model of Continual Learning with Cognitive Control

Russin¹,

Zolfaghar²,

Park³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…In addition, analogous to hippocampal place cells, which represent aspects of the physical environment 29 , 30 , hippocampal social-place-cells have been linked to identifying a person’s position in space 31 – 33 , which may facilitate spatial navigation 21 . The hippocampal-based cognitive mapping framework was recently supported by a study of Park et al showing that similar to physical space, social knowledge is encoded as a grid-like representation of a two-dimensional cognitive map mediated by the entorhinal cortex and hippocampus 34 . On the basis of this framework, one may speculate that the medial temporal lobe and associated networks in the frontal cortex and limbic system are responsible for encoding and altering the ‘position’ of a social counterpart within an internal map composed of the axes of power and affiliation.…”

Section: Discussionmentioning

confidence: 97%

Impaired remapping of social relationships in older adults

Oltmer

Wolbers

Kuehn

2021

Sci Rep

View full text Add to dashboard Cite

Social relationships are a central aspect of our everyday life, yet our ability to change established social relationships is an under-investigated topic. Here, we use the concept of cognitive mapping to investigate the plasticity of social relationships in younger and older adults. We describe social relationships within a ‘social space’, defined as a two-dimensional grid composed of the axis ‘power’ and ‘affiliation’, and investigate it using a 3D virtual environment with interacting avatars. We show that participants remap dimensions in ‘social space’ when avatars show conflicting behavior compared to consistent behavior and that, while older adults show similar updating behavior than younger adults, they show a distinct reduction in remapping social space. Our data provide first evidence that older adults show more rigid social behavior when avatars change their behavior in the dimensions of power and affiliation, which may explain age-related social behavior differences in everyday life.

show abstract

“…Like the hippocampus, however, grid cells are also not specifically tuned for space, as their properties have been described with computational principles ( Dordek et al, 2016 ; Stachenfeld et al, 2017 ; Baram et al, 2018 ; Behrens et al, 2018 ; Gershman, 2018 ; Momennejad and Howard, 2018 ; Bicanski and Burgess, 2019 ; Mark et al, 2020 ; Momennejad, 2020 ; Rueckemann et al, 2021 ), specifically, as a (basis) set of vectors that can linearly combine to represent any point in two dimensions. This “grid code” has since been found to facilitate non-spatial representation, such as for virtual ( Doeller et al, 2010 ), imagined ( Bellmund et al, 2016 ; Horner et al, 2016 ), conceptual ( Constantinescu et al, 2016 ), visual ( Julian et al, 2018 ; Nau et al, 2018 ), odor ( Bao et al, 2019 ), egocentric ( Moon et al, 2020 ), social ( Park et al, 2021 ), semantic ( Viganò et al, 2021 ), and contextual ( Julian and Doeller, 2021 ) information in humans.…”

Section: What Is a Cognitive Map?mentioning

confidence: 99%

“…Consider the recent study by Park et al (2021) , which also revealed that the grid code helps to make inferences. Participants learned pairs of 16 faces (AB) and how they compared along the dimension of popularity (e.g., A < B).…”

Section: What Is a Cognitive Map?mentioning

confidence: 99%

Spatialization of Time in the Entorhinal-Hippocampal System

Houser

2022

Front. Behav. Neurosci.

View full text Add to dashboard Cite

The functional role of the entorhinal-hippocampal system has been a long withstanding mystery. One key theory that has become most popular is that the entorhinal-hippocampal system represents space to facilitate navigation in one’s surroundings. In this Perspective article, I introduce a novel idea that undermines the inherent uniqueness of spatial information in favor of time driving entorhinal-hippocampal activity. Specifically, by spatializing events that occur in succession (i.e., across time), the entorhinal-hippocampal system is critical for all types of cognitive representations. I back up this argument with empirical evidence that hints at a role for the entorhinal-hippocampal system in non-spatial representation, and computational models of the logarithmic compression of time in the brain.

show abstract

Inferences on a multidimensional social hierarchy use a grid-like code

Cited by 131 publications

References 76 publications

A Neural Network Model of Continual Learning with Cognitive Control

A Neural Network Model of Continual Learning with Cognitive Control

Impaired remapping of social relationships in older adults

Spatialization of Time in the Entorhinal-Hippocampal System

Contact Info

Product

Resources

About