David Marr - Computational Investigation

Julesz developed a method for creating depth perception in images using random-dot patterns, contributing to understanding visual perception.

Focuses on how humans perceive the whole of something based on its parts, influencing the understanding of visual perception.

Demonstrated that mental rotation tasks involve mental imagery and that the time taken to mentally rotate an object correlates with its degree of rotation.

Barlow’s definition emphasizes the functional understanding of neurons in transmitting and responding to signals, crucial for understanding brain function.

Discovered specialized neurons for detecting complex objects like faces and hands, shedding light on neural processing of visual stimuli.

Emphasized the computational challenge in interpreting human vision, proposing that analyzing physical arrangements of surfaces could aid in interpreting images.

Introduced the Retinex theory, which explains color vision as a perceptual approximation to reflectance, influenced by surface properties like concavity or convexity.

Explored shape perception from shading, illumination, and surface geometry, contributing to understanding visual processing.

Demonstrates perceptual ambiguity and the brain’s ability to interpret 3D structures, highlighting the complexity of visual perception.

Emphasizes understanding the problem being solved over the mechanism used, crucial for comprehending complex systems like perception.

Focuses on perception as extracting information from the environment, emphasizing the role of invariants in perception and critiquing reliance on sense-data alone.

Addresses how vision selects relevant information from the environment, considering the purpose of vision in different animals.

Studies lesions and their effects on object recognition and perception, highlighting the distinction between recognizing an object’s shape and understanding its use.

Marr proposed a hierarchical model for understanding vision, emphasizing stages of representation from primal sketch to 3D models.