Is it possible to deeply understand neural representations through dimensionality reduction? Our recent work demonstrates that the visual representations in the human brain require interpretation within high-dimensional spaces. Moreover, we reveal that traditional methods like representational similarity analysis fail to detect this high-dimensional information in cortical activity; instead, a spectral approach is necessary. This research uncovers a vast expanse of uncharted dimensions that conventional techniques have overlooked but may be crucial for decoding the cortical code of human vision.
Raj Magesh Gauthaman, Brice Ménard, Michael F. Bonner
We examined whether visual neural networks align with brain representations due to shared constraints or universal features. Analysis of diverse networks revealed shared latent dimensions for image representation. Comparing these to human fMRI data showed brain-aligned representations are universal across networks. This suggests similarities between artificial and biological vision arise from core universal image representations learned convergently.
Zirui Chen, Michael F. Bonner
We find that untrained convolutional neural networks can produce brain-like visual representations. This challenges the view that extensive training is necessary for such similarities. The key factors are the networks’ architecture, specifically how they compress spatial information and expand feature information. This suggests that the basic structure of convolutional networks mimics biological vision constraints, allowing for cortex-like representations even without learning from experience.
Atlas Kazemian, Eric Elmoznino, Michael F. Bonner
Investigating deep neural networks (DNNs) as models for the visual cortex, we found that higher-dimensional representations in these networks better predict cortical responses and improve learning of new stimuli, challenging the idea that lower dimensionality enhances performance. This indicates that high-dimensional geometries might be advantageous for DNN models of visual processing.
Eric Elmoznino, Michael F. Bonner
PLOS Computational Biology (2023)
With technological advances allowing us to capture neural responses to thousands of stimuli across numerous channels (e.g., human fMRI, mouse two-photon imaging, monkey neuropixel probes), we face the challenge of analyzing vast, costly datasets. We must consider which computational tools are best suited for high-dimensional neural representation studies and what theoretical insights we can derive about neural representations from these large-scale datasets.
Raj Magesh Gauthaman, Florentin Guth, Atlas Kazemian, Zirui Chen, Michael F. Bonner
Cognitive Computational Neuroscience (2023)
Visit ^website for tutorial materials.
The lateral stream’s organization, thought to be hierarchical like in the ventral and dorsal streams, has not been thoroughly studied for naturalistic, social visuals. We compiled 250 3-second videos of everyday interactions between two people, annotated for visual features from low-level to high-level, including scene properties, social dynamics, and emotional content.
Emalie McMahon, Michael F. Bonner, Leyla Isik
What factors determine when visual memories will include details that go beyond perceptual experience? Here, seven experiments (N = 1,100 adults) explored whether spatial scale-specifically, perceived viewing distance-drives boundary extension. We created fake miniatures by exploiting tilt shift, a photographic effect that selectively reduces perceived distance while preserving other scene properties (e.g., making a distant railway appear like a model train). We found that visual memory is modulated by the spatial scale at which the environment is viewed.
Alon Hafri, Shreya Wadhwa, Michael F. Bonner
Using machine learning and fMRI, we tested if the human visual system encodes object co-occurrence statistics, revealed through individual object perception. We identified low-dimensional representations of these statistics in real-world scenes and correlated them with voxel-wise fMRI responses during object viewing. Our findings indicate cortical responses to single objects are influenced by their typical statistical contexts, highlighting brain regions where this connection is strongest.
Michael F. Bonner, Russell Epstein
Z Chen, MF Bonner. (2024) Universal dimensions of visual representation. arxiv
PDF
Code
A Kazemian, E Elmoznino, MF Bonner. (2024) Convolutional architectures are cortex-aligned de novo. bioRxiv
PDF
Code
Twitter
K Garcia, E McMahon, C Conwell, MF Bonner, L Isik. (2024) Modeling social vision highlights gaps between deep learning and humans.
bioRxiv
PDF
Twitter
Qu C, Bonner MF, DeWind NK, & Brannon EM (2024). Contextual coherence increases perceived numerosity independent of semantic content. PsyArXiv
PDF
Hafri AA, Bonner MF, Landau B, & Firestone C (2024). A phone in a basket looks like a knife in a cup: The perception of abstract relations. PsyArXiv
PDF
Twitter
Li SPD & Bonner MF (2023). Emergent selectivity for scenes, object properties, and contour statistics in feedforward models of scene-preferring cortex. bioRxiv
PDF
E Elmoznino, MF Bonner. (2024) High-performing neural network models of visual cortex benefit from high latent dimensionality. PLOS Computational Biology, 0(1): e1011792.
PDF
Code
BS Robinson, N Drenkow, C Conwell, MF Bonner. (2023) A sparse null code emerges in deep neural networks. NeurIPS UniReps Workshop
PDF
E McMahon, MF Bonner, L Isik. (2023) Hierarchical organization of social action features along the lateral visual pathway. Current Biology, 33, 1-13.
PDF
Code
C Magri, E Elmoznino, MF Bonner. (2023) Scene context is predictive of unconstrained object similarity judgments. Cognition, 239, 105535.
PDF
AA Hafri, S Wadhwa, MF Bonner. (2022) Perceived distance alters memory for scene boundaries. Psychological Science, 33(12), 2040–2058.
PDF
Code
F Lin, AA Hafri, MF Bonner. (2022) Scene memories are biased toward high-probability views. Journal of Experimental Psychology: Human Perception and Performance, 48(10): 1116-1129.
PDF
Code
A Harel, J Nador, MF Bonner, RA Epstein. (2022) Early electrophysiological markers of navigational affordances in scenes. Journal of Cognitive Neuroscience, 34(3): 397–410.
PDF
K Dwivedi, MF Bonner, RM Cichy, G Roig. (2021) Unveiling functions of the visual cortex using task-specific deep neural networks. PLOS Computational Biology, 17(8), e1009267.
PDF
MF Bonner, RA Epstein. (2021) Object representations in the human brain reflect the co-occurrence statistics of vision and language. Nature Communications, 12, 4081.
PDF
Code
Twitter
NK DeWind, MF Bonner, EM Brannon. (2020) Similarly oriented objects appear more numerous. Journal of Vision, 20, 4-4.
PDF
MF Bonner, RA Epstein. (2018) Computational mechanisms underlying cortical responses to the affordance properties of visual scenes. PLOS Computational Biology, 14:e1006111.
PDF
MF Bonner, RA Epstein. (2017) Coding of navigational affordances in the human visual system. Proceedings of the National Academy of Sciences, 114(18): 4793-8.
PDF
Code
AR Price, MF Bonner, JE Peelle, M Grossman. (2017) Neural coding of fine-grained object knowledge in perirhinal cortex. bioRxiv.
PDF
AR Price, JE Peelle, MF Bonner, M Grossman, RH Hamilton. (2016) Causal evidence for a mechanism of semantic integration in the angular gyrus revealed by high-definition transcranial direct current stimulation (HD-tDCS). Journal of Neuroscience, 36(13): 3829-38.
PDF
MF Bonner, AR Price, JE Peelle, M Grossman. (2016) Semantics of the visual environment encoded in parahippocampal cortex. Journal of Cognitive Neuroscience, 28(3): 361-78.
PDF
AR Price, MF Bonner, M Grossman. (2015) Semantic memory: cognitive and neuroanatomical perspectives. Brain Mapping: An Encyclopedic Reference. Toga AW, Poldrack RA (Eds.) Waltham: Academic Press. 529-36.
PDF
AR Price, MF Bonner, JE Peelle, M Grossman. (2015) Converging evidence for the neuroanatomic basis of combinatorial semantics in the angular gyrus. Journal of Neuroscience, 35:3276-84.
PDF
MF Bonner, M Grossman. (2014) The neural basis of semantic memory. Dementia and Memory. Nilsson L-G, Ohta N (Eds.) Hove: Psychology Press. 207-24.
MF Bonner, M Grossman. (2013) Deficits in semantic memory associated with focal neurodegenerative diseases. The Boston Process Approach to Neuropsychological Assessment: A Practitioner’s Guide. Ashendorf L, Swenson R, Libon DJ (Eds.) Oxford: Oxford University Press. 200-16.
MF Bonner, AR Price. (2013) Where is the anterior temporal lobe and what does it do? Journal of Neuroscience, 33:4213-5.
PDF
MF Bonner, JE Peelle, PA Cook, M Grossman. (2013) Heteromodal conceptual processing in the angular gyrus. NeuroImage, 71:175-86.
PDF
M Grossman, JE Peelle, EE Smith, CT McMillan, P Cook, J Powers, M Dreyfuss, MF Bonner, L Richmond, A Boller, E Camp, L Burkholder. (2013) Category-specific semantic memory: Converging evidence from BOLD fMRI and Alzheimer’s disease. NeuroImage, 68:263–74.
PDF
MF Bonner, M Grossman. (2012) Gray matter density of auditory association cortex relates to knowledge of sound concepts in primary progressive aphasia. Journal of Neuroscience, 32(23): 7986-91.
PDF
M Grossman, MF Bonner, J Weinstein. (2011) Music and Semantic Dementia–Reply. Archives of Neurology
PDF
J Weinstein, P Koenig, D Gunawardena, C McMillan, MF Bonner, M Grossman. (2011) Preserved musical semantic memory in semantic dementia. Archives of Neurology
PDF
MF Bonner, S Ash, M Grossman. (2010) The new classification of primary progressive aphasia into semantic, logopenic, or nonfluent/agrammatic variants. Current Neurology and Neuroscience Reports
PDF
C Farag, V Troiani, MF Bonner, C Powers, B Avants, J Gee, M Grossman. (2010) Hierarchical organization of scripts: converging evidence from fMRI and frontotemporal degeneration. Cerebral Cortex
PDF
MF Bonner, L Vesely, C Price, C Anderson, L Richmond, C Farag, B Avants, M Grossman. (2009) Reversal of the concreteness effect in semantic dementia. Cognitive Neuropsychology
PDF
L Vesely, MF Bonner, J Reilly, M Grossman. (2007) Free association in semantic dementia: The importance of being abstract. Brain and Language
R Zhang, ST Liu, W Chen, MF Bonner, J Pehrson, TJ Yen, PD Adams. (2007) HP1 proteins are essential for a dynamic nuclear response that rescues the function of perturbed heterochromatin in primary human cells. Molecular and Cellular Biology
PDF
© 2024 Bonner Lab.
Site made with Jekyll;
We are part of the Kreiger School of Art and Science at Johns Hopkins University
