A Rich Source of Labels for Deep Network Models of the Primate Dorsal Visual Stream

Deep convolutional neural networks (CNNs) have structures that are loosely related to that of the primate visual cortex. Surprisingly, when these networks are trained for object classification, the activity of their early, intermediate, and later layers becomes closely related to activity patterns in corresponding parts of the primate ventral visual stream. The activity statistics are far from identical, but perhaps remaining differences can be minimized in order to produce artificial networks with highly brain-like activity and performance, which would provide a rich source of insight into primate vision. One way to align CNN activity more closely with neural activity is to add cost functions that directly drive deep layers to approximate neural recordings. However, suitably large datasets are particularly difficult to obtain for deep structures, such as the primate middle temporal area (MT). To work around this barrier, we have developed a rich empirical model of activity in MT. The model is pixel-computable, so it can provide an arbitrarily large (but approximate) set of labels to better guide learning in the corresponding layers of deep networks. Our model approximates a number of MT phenomena more closely than previous models. Furthermore, our model approximates population statistics in detail through fourteen parameter distributions that we estimated from the electrophysiology literature. In general, deep networks with internal representations that closely approximate those of the brain may help to clarify the mechanisms that produce these representations, and the roles of various properties of these representations in performance of vision tasks. Although our empirical model inevitably differs from real neural activity, it allows tuning properties to be modulated independently, which may allow very detailed exploration of the origins and functional roles of these properties.

[1]  Wyeth Bair,et al.  A Model of Binocular Motion Integration in MT Neurons , 2016, The Journal of Neuroscience.

[2]  Jian Sun,et al.  Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[3]  James J. DiCarlo,et al.  How Does the Brain Solve Visual Object Recognition? , 2012, Neuron.

[4]  Honglak Lee,et al.  Sparse deep belief net model for visual area V2 , 2007, NIPS.

[5]  Daniel P. Huttenlocher,et al.  Efficient Belief Propagation for Early Vision , 2004, Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004..

[6]  H. Tamura,et al.  Visual response properties of cells in the ventral and dorsal parts of the macaque inferotemporal cortex. , 2001, Cerebral cortex.

[7]  Daniel B. Rubin,et al.  The Stabilized Supralinear Network: A Unifying Circuit Motif Underlying Multi-Input Integration in Sensory Cortex , 2015, Neuron.

[8]  Sinan Kalkan,et al.  Deep Hierarchies in the Primate Visual Cortex: What Can We Learn for Computer Vision? , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  RussLL L. Ds Vnlos,et al.  SPATIAL FREQUENCY SELECTIVITY OF CELLS IN MACAQUE VISUAL CORTEX , 2022 .

[10]  E. Peli Contrast in complex images. , 1990, Journal of the Optical Society of America. A, Optics and image science.

[11]  J. Movshon,et al.  Dynamics of motion signaling by neurons in macaque area MT , 2005, Nature Neuroscience.

[12]  J. Gallant,et al.  A Three-Dimensional Spatiotemporal Receptive Field Model Explains Responses of Area MT Neurons to Naturalistic Movies , 2011, The Journal of Neuroscience.

[13]  Lawrence C. Sincich,et al.  Bypassing V1: a direct geniculate input to area MT , 2004, Nature Neuroscience.

[14]  J Anthony Movshon,et al.  Properties of pattern and component direction-selective cells in area MT of the macaque. , 2016, Journal of neurophysiology.

[15]  Daniel L. K. Yamins,et al.  Deep Neural Networks Rival the Representation of Primate IT Cortex for Core Visual Object Recognition , 2014, PLoS Comput. Biol..

[16]  Nikolaus Kriegeskorte,et al.  Deep Supervised, but Not Unsupervised, Models May Explain IT Cortical Representation , 2014, PLoS Comput. Biol..

[17]  Ricardo Gattass,et al.  Electrophysiological Imaging of Functional Architecture in the Cortical Middle Temporal Visual Area of Cebus apella Monkey , 2003, The Journal of Neuroscience.

[18]  Douglas J. Bakkum,et al.  Revealing neuronal function through microelectrode array recordings , 2015, Front. Neurosci..

[19]  Eero P. Simoncelli,et al.  How MT cells analyze the motion of visual patterns , 2006, Nature Neuroscience.

[20]  Julien Marzat,et al.  Real-Time Dense and Accurate Parallel Optical Flow using CUDA , 2009 .

[21]  James G. King,et al.  Reconstruction and Simulation of Neocortical Microcircuitry , 2015, Cell.

[22]  Ha Hong,et al.  Explicit information for category-orthogonal object properties increases along the ventral stream , 2016, Nature Neuroscience.

[23]  John H. R. Maunsell,et al.  Topographic organization of the middle temporal visual area in the macaque monkey: Representational biases and the relationship to callosal connections and myeloarchitectonic boundaries , 1987, The Journal of comparative neurology.

[24]  P. J. Green,et al.  Density Estimation for Statistics and Data Analysis , 1987 .

[25]  Christopher C. Pack,et al.  Temporal dynamics of a neural solution to the aperture problem in visual area MT of macaque brain , 2001, Nature.

[26]  M. Carandini,et al.  Normalization as a canonical neural computation , 2011, Nature Reviews Neuroscience.

[27]  Eero P. Simoncelli,et al.  A model of neuronal responses in visual area MT , 1998, Vision Research.

[28]  T. Albright,et al.  Adaptive Surround Modulation in Cortical Area MT , 2007, Neuron.

[29]  G. Orban,et al.  The spatial distribution of the antagonistic surround of MT/V5 neurons. , 1997, Cerebral cortex.

[30]  W. Wildman,et al.  Theoretical Neuroscience , 2014 .

[31]  S. Treue,et al.  Attentional Modulation Strength in Cortical Area MT Depends on Stimulus Contrast , 2002, Neuron.

[32]  Bartlett W. Mel,et al.  Computational subunits in thin dendrites of pyramidal cells , 2004, Nature Neuroscience.

[33]  Bart G Borghuis,et al.  Temporal dynamics of direction tuning in motion-sensitive macaque area MT. , 2005, Journal of neurophysiology.

[34]  Thaddeus B. Czuba,et al.  Area MT Encodes Three-Dimensional Motion , 2014, The Journal of Neuroscience.

[35]  R. Andersen,et al.  Center–Surround Antagonism Based on Disparity in Primate Area MT , 1998, The Journal of Neuroscience.

[36]  Rob Fergus,et al.  Visualizing and Understanding Convolutional Networks , 2013, ECCV.

[37]  Fritjof Helmchen,et al.  Chronic imaging of cortical sensory map dynamics using a genetically encoded calcium indicator , 2012, The Journal of physiology.

[38]  D. Bradley,et al.  Structure and function of visual area MT. , 2005, Annual review of neuroscience.

[39]  Nikola T. Markov,et al.  A Weighted and Directed Interareal Connectivity Matrix for Macaque Cerebral Cortex , 2012, Cerebral cortex.

[40]  H. Rodman,et al.  Coding of visual stimulus velocity in area MT of the macaque , 1987, Vision Research.

[41]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[42]  William T Newsome,et al.  Middle Temporal Visual Area Microstimulation Influences Veridical Judgments of Motion Direction , 2002, The Journal of Neuroscience.

[43]  W. Newsome,et al.  A selective impairment of motion perception following lesions of the middle temporal visual area (MT) , 1988, The Journal of neuroscience : the official journal of the Society for Neuroscience.

[44]  Thomas Brox,et al.  A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[45]  Takeo Kanade,et al.  An Iterative Image Registration Technique with an Application to Stereo Vision , 1981, IJCAI.

[46]  Nicholas J. Priebe,et al.  Tuning for Spatiotemporal Frequency and Speed in Directionally Selective Neurons of Macaque Striate Cortex , 2006, The Journal of Neuroscience.

[47]  Guy A Orban,et al.  Higher order visual processing in macaque extrastriate cortex. , 2008, Physiological reviews.

[48]  C. D. Kemp,et al.  Density Estimation for Statistics and Data Analysis , 1987 .

[49]  W. Newsome,et al.  Deficits in visual motion processing following ibotenic acid lesions of the middle temporal visual area of the macaque monkey , 1985, The Journal of neuroscience : the official journal of the Society for Neuroscience.

[50]  G. DeAngelis,et al.  A Logarithmic, Scale-Invariant Representation of Speed in Macaque Middle Temporal Area Accounts for Speed Discrimination Performance , 2005, The Journal of Neuroscience.

[51]  J. Movshon,et al.  Motion Integration by Neurons in Macaque MT Is Local, Not Global , 2007, The Journal of Neuroscience.

[52]  H. Hirschmüller Accurate and Efficient Stereo Processing by Semi-Global Matching and Mutual Information , 2005, CVPR.

[53]  T. Pasternak,et al.  Transient and permanent deficits in motion perception after lesions of cortical areas MT and MST in the macaque monkey. , 1999, Cerebral cortex.

[54]  Bryan P. Tripp,et al.  Decorrelation of Spiking Variability and Improved Information Transfer Through Feedforward Divisive Normalization , 2012, Neural Computation.

[55]  A. Fairhall,et al.  Fractional differentiation by neocortical pyramidal neurons , 2008, Nature Neuroscience.

[56]  D C Van Essen,et al.  Functional properties of neurons in middle temporal visual area of the macaque monkey. I. Selectivity for stimulus direction, speed, and orientation. , 1983, Journal of neurophysiology.

[57]  E. Parzen On Estimation of a Probability Density Function and Mode , 1962 .

[58]  Yoshua Bengio,et al.  Show, Attend and Tell: Neural Image Caption Generation with Visual Attention , 2015, ICML.

[59]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[60]  S. Treue,et al.  Misperceptions of speed are accounted for by the responses of neurons in macaque cortical area MT. , 2011, Journal of neurophysiology.

[61]  Ha Hong,et al.  Performance-optimized hierarchical models predict neural responses in higher visual cortex , 2014, Proceedings of the National Academy of Sciences.

[62]  J. Robson Spatial and Temporal Contrast-Sensitivity Functions of the Visual System , 1966 .

[63]  Stefan Treue,et al.  Feature-based attention influences motion processing gain in macaque visual cortex , 1999, Nature.

[64]  D. J. Felleman,et al.  Distributed hierarchical processing in the primate cerebral cortex. , 1991, Cerebral cortex.

[65]  H. Akaike A new look at the statistical model identification , 1974 .

[66]  Ali Borji,et al.  State-of-the-Art in Visual Attention Modeling , 2013, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[67]  J. Movshon,et al.  Adaptation changes the direction tuning of macaque MT neurons , 2004, Nature Neuroscience.

[68]  Andreas T. Schaefer,et al.  Coincidence detection in pyramidal neurons is tuned by their dendritic branching pattern. , 2003, Journal of neurophysiology.

[69]  Yuwei Cui,et al.  Diverse Suppressive Influences in Area MT and Selectivity to Complex Motion Features , 2013, The Journal of Neuroscience.

[70]  Gregory C DeAngelis,et al.  Coding of horizontal disparity and velocity by MT neurons in the alert macaque. , 2003, Journal of neurophysiology.

[71]  Marcel van Gerven,et al.  Increasingly complex representations of natural movies across the dorsal stream are shared between subjects , 2017, NeuroImage.

[72]  Christopher C. Pack,et al.  Contrast dependence of suppressive influences in cortical area MT of alert macaque. , 2005, Journal of neurophysiology.