Bradley C. Love

Distinct hippocampal mechanisms support concept formation and updating

The paper uses fMRI and computational models to identify neural mechanisms for memory creation and updating during learning, finding distinct hippocampal involvement in learning outcomes.

Adaptive stretching of representations across brain regions and deep learning model layers

The paper shows that brain regions and deep learning models stretch representations along task-relevant dimensions to optimize performance through error minimization.

The inevitability and superfluousness of cell types in spatial cognition

The paper argues that spatial cell types in the brain may not be essential for spatial cognition, as similar patterns emerge in non-spatial deep neural networks.

A controller-peripheral architecture and costly energy principle for learning

The paper proposes a controller-peripheral architecture to model brain coordination for learning, showing how higher-level control shapes lower-level representations and outperforms gradient descent models.

A multi-level account of hippocampal function from behavior to neurons

The paper decomposes a cognitive model into neuron-like units to explain how neural populations coordinate to form cognitive representations.

A neural network account of memory replay and knowledge consolidation

The paper uses a neural network to simulate memory replay and consolidation, showing generative replay enhances generalization and supports adaptive learning.

Directly interfacing brain and deep networks exposes non-hierarchical visual processing

The paper tests if brain regions correspond to deep network layers by replacing model activity with brain activity, finding that all ventral visual stream regions best match later network layers.

Abstract neural representations of category membership beyond information coding stimulus or response

The paper investigates whether the brain forms abstract category representations independent of sensory or motor details using fMRI and cognitive modeling.

Subjective value and decision entropy are jointly encoded by aligned gradients across the human brain

The paper shows that subjective value and decision entropy are encoded by aligned gradients across the human brain, with value areas more strongly signaling decision entropy than value itself.

Ventromedial prefrontal cortex compression during concept learning

This study shows that neural compression in the ventromedial prefrontal cortex during learning predicts the ability to focus on relevant information.

A non-spatial account of place and grid cells based on clustering models of concept learning

This paper shows that a clustering model of concept learning can explain both spatial navigation and conceptual knowledge by generating place and grid cell-like representations in spatial tasks and more abstract representations in conceptual tasks.

Measures of neural similarity

The paper evaluates neural similarity measures across fMRI datasets, finding that alternatives to Pearson correlation outperform it in capturing similarity across tasks.

Levels of Representation in a Deep Learning Model of Categorization

This paper investigates how deep learning models of categorization use different levels of representation to inform memory and learning, showing that higher layers capture more complex features and influence categorization biases.

Dynamic integration of conceptual information during learning

The paper investigates how conceptual representations are formed during learning using a combination of behavioral experiments and computational modeling.

Occipitotemporal Representations Reflect Individual Differences in Conceptual Knowledge

The paper shows that occipitotemporal cortex representations reflect individual differences in conceptual knowledge through attentional weighting of visual features.

Estimating the functional dimensionality of neural representations

This paper develops a method to estimate the functional dimensionality of neural representations in fMRI data, showing it accurately recovers true dimensionality in simulations and highlights applications in evaluating models and understanding brain function.