Emergence of brain-like mirror-symmetric viewpoint tuning in convolutional neural networks
- Rockefeller University, United States
- Columbia University, United States
Abstract
Primates can recognize objects despite 3D geometric variations such as in-depth rotations. The computational mechanisms that give rise to such invariances are yet to be fully understood. A curious case of partial invariance occurs in the macaque face-patch AL and in fully connected layers of deep convolutional networks in which neurons respond similarly to mirror-symmetric view (e.g., left and right profiles). Why does this tuning develop? Here, we propose a simple learning-driven explanation for mirror-symmetric viewpoint tuning. We show that mirror-symmetric viewpoint tuning for faces emerges in the fully connected layers of convolutional deep neural networks trained on object recognition tasks, even when the training dataset does not include faces. First, using 3D objects rendered from multiple views as test stimuli, we demonstrate that mirror-symmetric viewpoint tuning in convolutional neural network models is not unique to faces: it emerges for multiple object categories with bilateral symmetry. Second, we show why this invariance emerges in the models. Learning to discriminate among bilaterally symmetric object categories induces reflection-equivariant intermediate representations. AL-like mirror-symmetric tuning is achieved when such equivariant responses are spatially pooled by downstream units with sufficiently large receptive fields. These results explain how mirror-symmetric viewpoint tuning can emerge in neural networks, providing a theory of how they might emerge in the primate brain. Our theory predicts that mirror-symmetric viewpoint tuning can emerge as a consequence of exposure to bilaterally symmetric objects beyond the category of faces, and that it can generalize beyond previously experienced object categories.
Data availability
The stimulus set and the source code required for reproducing our results are available at https://gitfront.io/r/afarzmahdi/p666tmWy7YuY/AL-symmetry-manuscript-codes/.
-
Object categories across viewshttps://github.com/amirfarzmahdi/AL-Symmetry.
Article and author information
Author details
Amirhossein Farzmahdi
Winrich A Freiwald
Nikolaus Kriegeskorte
Funding
National Eye Institute (R01EY021594)
- Winrich A Freiwald
National Eye Institute (R01EY029998)
- Winrich A Freiwald
National Institute of Neurological Disorders and Stroke (RF1NS128897)
- Nikolaus Kriegeskorte
Naval Research Laboratory (N00014-20-1-2292)
- Winrich A Freiwald
Charles H. Revson Foundation
- Tal Golan
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Copyright
© 2024, Farzmahdi et al.
This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 834
- views
-
- 141
- downloads
-
- 0
- citations
Views, downloads and citations are aggregated across all versions of this paper published by eLife.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Emergence of brain-like mirror-symmetric viewpoint tuning in convolutional neural networks
eLife 13:e90256.
https://doi.org/10.7554/eLife.90256
Further reading
-
- Neuroscience
Multiplexed error-robust fluorescence in situ hybridization (MERFISH) allows genome-scale imaging of RNAs in individual cells in intact tissues. To date, MERFISH has been applied to image thin-tissue samples of ~10 µm thickness. Here, we present a thick-tissue three-dimensional (3D) MERFISH imaging method, which uses confocal microscopy for optical sectioning, deep learning for increasing imaging speed and quality, as well as sample preparation and imaging protocol optimized for thick samples. We demonstrated 3D MERFISH on mouse brain tissue sections of up to 200 µm thickness with high detection efficiency and accuracy. We anticipate that 3D thick-tissue MERFISH imaging will broaden the scope of questions that can be addressed by spatial genomics.
-
- Neuroscience
Learning alters cortical representations and improves perception. Apical tuft dendrites in cortical layer 1, which are unique in their connectivity and biophysical properties, may be a key site of learning-induced plasticity. We used both two-photon and SCAPE microscopy to longitudinally track tuft-wide calcium spikes in apical dendrites of layer 5 pyramidal neurons in barrel cortex as mice learned a tactile behavior. Mice were trained to discriminate two orthogonal directions of whisker stimulation. Reinforcement learning, but not repeated stimulus exposure, enhanced tuft selectivity for both directions equally, even though only one was associated with reward. Selective tufts emerged from initially unresponsive or low-selectivity populations. Animal movement and choice did not account for changes in stimulus selectivity. Enhanced selectivity persisted even after rewards were removed and animals ceased performing the task. We conclude that learning produces long-lasting realignment of apical dendrite tuft responses to behaviorally relevant dimensions of a task.
