Contents

• Introduction
  • Receptive fields in the visual pathway: basic issues
    • Detailed profiles
    • Spatial acuity
    • Function
  • Parallel distributed processing and RFs
    • General-purpose computing with connections
    • Computing with structured connection patterns
    • Convolutional NNs and receptive fields
• Sensory coding
  • Interpolation or channel coding?
  • Feature detectors or spatial frequency analyzers?
    • Gratings
    • Features
  • Suspicious coincidences and Barlow's Probabilistic Pandemonium
  • Statistics of natural images
• Recognition
  • Remapping of similarity spaces
    • Inverse optics: RFs and regularization
    • Similarity within and between object classes
  • Recognition as function approximation
• Categorization
  • Similarity to prototypes
  • Conditions for veridical representation
  • Constraints on the distal structure
  • Constraints on the distal-to-proximal mapping
    • Strict rank preservation.
    • Local rank preservation.
    • Local approximate rank preservation.
• Summary
  • The new Pandemonium
  • Conclusion
  • Acknowledgments
• References
• About this document ...