1
views
0
recommends
+1 Recommend
0 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      'Artiphysiology' reveals V4-like shape tuning in a deep network trained for image classification

      research-article

      Read this article at

      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          Abstract

          Deep networks provide a potentially rich interconnection between neuroscientific and artificial approaches to understanding visual intelligence, but the relationship between artificial and neural representations of complex visual form has not been elucidated at the level of single-unit selectivity. Taking the approach of an electrophysiologist to characterizing single CNN units, we found many units exhibit translation-invariant boundary curvature selectivity approaching that of exemplar neurons in the primate mid-level visual area V4. For some V4-like units, particularly in middle layers, the natural images that drove them best were qualitatively consistent with selectivity for object boundaries. Our results identify a novel image-computable model for V4 boundary curvature selectivity and suggest that such a representation may begin to emerge within an artificial network trained for image categorization, even though boundary information was not provided during training. This raises the possibility that single-unit selectivity in CNNs will become a guide for understanding sensory cortex.

          Related collections

          Most cited references36

          • Record: found
          • Abstract: found
          • Article: not found

          Selectivity and tolerance ("invariance") both increase as visual information propagates from cortical area V4 to IT.

          Our ability to recognize objects despite large changes in position, size, and context is achieved through computations that are thought to increase both the shape selectivity and the tolerance ("invariance") of the visual representation at successive stages of the ventral pathway [visual cortical areas V1, V2, and V4 and inferior temporal cortex (IT)]. However, these ideas have proven difficult to test. Here, we consider how well population activity patterns at two stages of the ventral stream (V4 and IT) discriminate between, and generalize across, different images. We found that both V4 and IT encode natural images with similar fidelity, whereas the IT population is much more sensitive to controlled, statistical scrambling of those images. Scrambling sensitivity was proportional to receptive field (RF) size in both V4 and IT, suggesting that, on average, the number of visual feature conjunctions implemented by a V4 or IT neuron is directly related to its RF size. We also found that the IT population could better discriminate between objects across changes in position, scale, and context, thus directly demonstrating a V4-to-IT gain in tolerance. This tolerance gain could be accounted for by both a decrease in single-unit sensitivity to identity-preserving transformations (e.g., an increase in RF size) and an increase in the maintenance of rank-order object selectivity within the RF. These results demonstrate that, as visual information travels from V4 to IT, the population representation is reformatted to become more selective for feature conjunctions and more tolerant to identity preserving transformations, and they reveal the single-unit response properties that underlie that reformatting.
            Bookmark
            • Record: found
            • Abstract: found
            • Article: not found

            Responses to contour features in macaque area V4.

            The ventral pathway in visual cortex is responsible for the perception of shape. Area V4 is an important intermediate stage in this pathway, and provides the major input to the final stages in inferotemporal cortex. The role of V4 in processing shape information is not yet clear. We studied V4 responses to contour features (angles and curves), which many theorists have proposed as intermediate shape primitives. We used a large parametric set of contour features to test the responses of 152 V4 cells in two awake macaque monkeys. Most cells responded better to contour features than to edges or bars, and about one-third exhibited systematic tuning for contour features. In particular, many cells were selective for contour feature orientation, responding to angles and curves pointing in a particular direction. There was a strong bias toward convex (as opposed to concave) features, implying a neural basis for the well-known perceptual dominance of convexity. Our results suggest that V4 processes information about contour features as a step toward complex shape recognition.
              Bookmark
              • Record: found
              • Abstract: found
              • Article: found
              Is Open Access

              Deep Neural Networks Reveal a Gradient in the Complexity of Neural Representations across the Brain's Ventral Visual Pathway

              Converging evidence suggests that the mammalian ventral visual pathway encodes increasingly complex stimulus features in downstream areas. Using deep convolutional neural networks, we can now quantitatively demonstrate that there is indeed an explicit gradient for feature complexity in the ventral pathway of the human brain. Our approach also allows stimulus features of increasing complexity to be mapped across the human brain, providing an automated approach to probing how representations are mapped across the cortical sheet. Finally, it is shown that deep convolutional neural networks allow decoding of representations in the human brain at a previously unattainable degree of accuracy, providing a more sensitive window into the human brain.
                Bookmark

                Author and article information

                Contributors
                Role: Reviewing Editor
                Role: Senior Editor
                Journal
                eLife
                Elife
                eLife
                eLife
                eLife Sciences Publications, Ltd
                2050-084X
                20 December 2018
                2018
                : 7
                Affiliations
                [1 ]deptDepartment of Biological Structure, Washington National Primate Research Center University of Washington SeattleUnited States
                [2 ]University of Washington Institute for Neuroengineering SeattleUnited States
                [3 ]deptComputational Neuroscience Center University of Washington SeattleUnited States
                The Hebrew University of Jerusalem Israel
                University of Pennsylvania United States
                The Hebrew University of Jerusalem Israel
                Article
                38242
                10.7554/eLife.38242
                6335056
                30570484
                6a5e1d79-e6e9-43da-ade7-a9fba3399acd
                © 2018, Pospisil et al

                This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

                Product
                Funding
                Funded by: FundRef http://dx.doi.org/10.13039/100000001, National Science Foundation;
                Award ID: Graduate Research Fellowship
                Award Recipient :
                Funded by: FundRef http://dx.doi.org/10.13039/100000001, National Science Foundation;
                Award ID: CRCNS Grant IIS-1309725
                Award Recipient :
                Funded by: FundRef http://dx.doi.org/10.13039/100006785, Google;
                Award ID: Google Faculty Research Award
                Award Recipient :
                Funded by: FundRef http://dx.doi.org/10.13039/100000002, National Institutes of Health;
                Award ID: Grant R01 EY-018839
                Award Recipient :
                Funded by: NIH Office of Research Infrastructure Programs;
                Award ID: Grant RR-00166 to the Washington National Primate Research Center
                Award Recipient :
                The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
                Categories
                Research Article
                Neuroscience
                Custom metadata
                Single units in a deep convolutional neural network trained for image classification develop shape selectivity that is similar to that found in the primate visual cortex.

                Life sciences
                neural network,v4,shape perception,ventral stream,rhesus macaque
                Life sciences
                neural network, v4, shape perception, ventral stream, rhesus macaque

                Comments

                Comment on this article