Perception

Modified: 2026-01-25 10:23 PM

Sensation and Perception are among the oldest topics in psychology. Gibson argued for direct perception. Marr suggested a multistep system for analyzing the visual world. Milner and Goodale sought to discover the neural pathways related to perception and action.

Chapter 3 Perception: Beyond Gibson's (1950) Direct Perception--Vicki Bruce and Yoav Tamor

Background of the Classic Study
- James Gibson
- Ecological Psychology
- Studied driver behavior first
- Retinal image loses third dimension
- Perspective in art
- FYI, the "sleepers" on page 22 are railroad ties
- Motion parallax: objects moving at the same speed appear to move faster when close by
  - Example: B-52 flying at 30,000 ft. appears to move slowly in the sky because it is far away; crop duster flying just overhead appears to move very fast (but the B-52 is faster)
- Landing an airplane involves depth vision cues
- Gibson's book, The Perception of the Visual World (1950) (review) divided researcherss
Detailed Description of the Classic Study
- Gibson's 'Classic' Contributions: Texture Gradients and Optic Flow
  - Affordances
  - Example: Most chairs are around 17 inches tall at the seat. Now think of NBA basketball players. Do they need a higher seat?
    - Would your knees be as high as his on the same chair?

Direct realism of perception, no interpretation needed
Gibson did not deny:
- Physiological processes
- Importance of image formation
- He did deny the idea that the retinal image be the starting point of perception
Texture Gradients
- Importance of the visual background
- Background consists of: surfaces, textures, texture gradients. Those provide a scale for perceiving distance and orientation
  - Such analyses "allow a completely different take on the classic problem of depth perception. (p. 25)
- Gibson on size constancy: at any distance, the same object will occude the same amount of background
- Size constancy can break down when looking from great heights
- Ames Room
Optic Flow
- Inflow and Outflow

Impact of the Classic Study
- Video: David Lee's swinging room
- tau (see p. 29 on gannets)
  - Video: Gannets diving
- Gibson and optic flow:
  - Natural scenes
    - Primate visual tuning for distinguishing ripe fruit from foliage
  - Natural activities
    - Eye movement analysis for where and when
  - Ecological psychology
    - Invariants
    - Artificiality of the laboratory
    - Video: Baseball batting and information processing vs ecological theory
      - Longish video but very interesting. NOT ON THE TEST but will help you
Critique of the Classic Study
- tau not proven
- Optical flow plus binocular disparity
- Flexible use of information
- Marr:
  - Detection of physical invariants is an information processing problem
  - Underated the sheer difficulty of such detection
Conclusions
- Moving from laboratory to natural world was important
- Visual-Neuro-Ecological awareness
  - Meaning: Niches are different for different species in same locations
- Biologically inspired robotics
- Gibson and Skinner on Embodied Robots (Abstract)
- Summary of Gibson's ideass
- Video: Direct Perception

Chapter 4 Perception: Beyond Marr's (1982) Computational Approach to Vision--George Mather

Background of the Classic Study
- David Marr
  - Vision (2010)--Full text of Marr's book
- Hubel and Wiesel
- Feature Detectors
- Apple II and IBM PC
  - Both debuted when researchers were developing computational approaches
Detailed Description of the Classic Study
- Marr's Thesis
  - General framework for studying vision
  - Comprehensive theory of visual processing
- Marr's Framework (levels are largely independent of each other)
  - Computational Theory
    - Constraints: Use-once and Continuity
  - Representations and Algorithms
  - Hardware Implementation
  - Summary: Identify constraints, develop algorithms, implement by different hardware
- Marr's Theory of Vision
  - Lisp (Computer language)
  - Bottom-Up Theory
  - Three parts:
    - Primal sketch
      - Visual primitives: edges, bars, blobs, terminations
        
        Similar to Schank's primitives (from Michael Brenndoerfer)
        
        Conceptual Dependency solved the meaning representation problem by introducing a small set of primitive actions that could be combined to represent any event. These primitives were chosen to be at the right level of abstraction: specific enough to distinguish different types of events, but general enough to capture semantic equivalence across diverse surface forms. The theory proposed approximately a dozen primitive actions, including:
        
        ATRANS: Transfer of abstract relationships (possession, ownership, control)
        
        PTRANS: Transfer of physical location (movement of objects)
        
        PROPEL: Application of physical force
        
        MOVE: Movement of a body part
        
        GRASP: Physical grasping of an object
        
        INGEST: Taking something into the body
        
        EXPEL: Removing something from the body
        
        MTRANS: Transfer of mental information
        
        MBUILD: Construction of new thoughts from old ones
        
        SPEAK: Production of sounds
        
        ATTEND: Focusing sense organs on a stimulus
        
        Schank used these primitives to describe common social situations: eating at restaurant, attending a birthday party
    - 2-1/2 D Sketch
      - Creates dense depth map from visual primitives including: local surface orientation, disparity, motion, shading, texture, contour
    - 3-D Model Representation
      - Describes shapes using primitives of volume and shape
      - Representation is independent of earlier stages and is akin to pipe models or balloon animals
      - Later matched with items from stored catalog of 3-D models
    Marr pipecleaner models (from ResearchGate)
Impact and Critique of the Classic Study
- Levels of Understanding
  - Laid foundations for Computational Neuroscience
  - Marr and Poggio's (1976) original levels: computation, algorithm, mechanism, hardware implementation
  - Poggio's (2012) levels: evolution, learning and development, computation, algorithms, implementation (including wetware and hardware)
  - Computational level theories all have the status of hypotheses
- Theory of Processing
  - Top-Down and Bottom-Up processing occur in vision
    - Recent research (Pouget, Dayan, & Zemel, 2000) suggests that visual measurements are represented continously in large populations of cells
    - Alternatives to Marr's theory exist
    - Marr's analysis of motion detection is not supported
    - New analyses of stereopsis exist (including constraints and algorithms not considered by Marr)
    - Storage requirements for Marr's representations may be too onerous
  - Modern artificial neural networks have been spectacularly successful thanks to:
    - Greater processing power
    - Machine learning
    - Massive databases containing images and videos
Conclusions
- Levels of understanding framework longest lasting piece of theory
  - Adequate theories of perception must:
    - Incorporate computational theory
    - Align theory to algorithms and potential neural substrates
- Marr's approach not usually taught or adopted as a methodology
- Computational theories can be mathematically plausible but may not reflect organismic realities

Chapter 5 Perception: Beyond Goodale and Milner's (1992) Separate Visual Pathways--Glyn W. Humphreys and Edward de Haan

Background and Description of the Classic Study
- David Milner
- Melvyn Goodale
- Two Streams Hypothesis
  - Ventral--"what" stream
  - Dorsal--"where" stream (later "how" stream)
- Patient DF
Perception and Action: The Case from Neuropsychology
- Dorsal stream--online, real time
- Ventral stream--slower
Additional Properties of the Ventral and Dorsal Streams
- Task difficulty, no general effect
- Milner and Goodale (1995):
  - Ventral stream--offline, delayed reaching and grasping
  - Dorsal stream--online guidance of reaching and grasping, no support for delayed action
Converging Evidence from Normal Participants
- Looking at normal perception and action
  - Ventral--conscious perceptual judgments
  - Dorsal--operates unconsciously
- Visual Illusions
  - Ebbinghaus Illusion
    - Conscious perceptual judgments affected by context
    - Grasp appertures not affected
Impact of the Classic Study
- Milner and Goodale (1995) had major impact including on:
  - somatosensory
  - auditory
  - speech perception
  - visual extinction
  - visual neglect
- Di Pellegrino, Rafal, and Tipper (2005, Abstract)
  - "This contralesional extinction was significantly reduced when cups had handles affording a left-hand grasp, even though no hand response was required."
- Susceptibility to Illusions

Critique of the Classic Study

See Table from Norman J. (2002). "Two visual systems and two theories of perception: An attempt to reconcile the constructivist and ecological approaches". Behav Brain Sci. 25 (1): 73–144. doi:10.1017/s0140525x0200002x. PMID 12625088. Compare to Table 5.1 on

Factor Ventral system (what) Dorsal system (how)

Function Recognition/identification Visually guided behaviour

Sensitivity High spatial frequencies - details High temporal frequencies - motion

Memory Long-term stored representations Only very short-term storage

Speed Relatively slow Relatively fast

Consciousness Typically high Typically low

Frame of reference Allocentric or object-centered Egocentric or viewer-centered

Visual input Mainly foveal or parafoveal Across retina

Monocular vision Generally reasonably small effects Often large effects e.g. motion parallax

Caveat 1: DF

DF suffers from optic ataxia as well

DF does not present perfect dissociation

DF suffered CO poisoning

Caveat 2: Optic Ataxia

Less optic ataxia with familiar objects

Less optic ataxia for peripheral targets

Optic ataxia does not represent spared perception

Caveat 3: Two Anatomically Separate and Functionally Independent Streams

Ventral and Dorsal streams not anatomically separate

There is functional crosstalk between the two streams

Ventral and Dorsal are not the only neural pathways

Caveat 4: Illusions and Actions

Ventral and Dorsal streams only not sufficient to explain illusions

Actions, too, can be influenced by illusions

Caveat 5: Perception in the Dorsal Stream and Action in the Ventral

Functional Brain Imaging

Picture is more complex than object recognition mediated by ventral stream and reaching and grasping by dorsal stream

Glass patterns Lestou (2014)

Normal patients: patterned vs random--ventral and dorsal activated

Patient with bilateral lesions of ventral cortex--normal activity in dorsal region

Patient with optic ataxia with dorsal brain lesions--poor at detecting patterns, did not discriminate between random and patterns

So, dorsal cortex is necessary in pattern recognition within ventral visual system (see p. 59)

Factor	Ventral system (what)	Dorsal system (how)
Function	Recognition/identification	Visually guided behaviour
Sensitivity	High spatial frequencies - details	High temporal frequencies - motion
Memory	Long-term stored representations	Only very short-term storage
Speed	Relatively slow	Relatively fast
Consciousness	Typically high	Typically low
Frame of reference	Allocentric or object-centered	Egocentric or viewer-centered
Visual input	Mainly foveal or parafoveal	Across retina
Monocular vision	Generally reasonably small effects	Often large effects e.g. motion parallax

Action Coding (Humphreys, 2010)
- Participants had increased activity in ventral visual stream when the objects apppeared to interact
  - Not learned, an affordance
  - Information in the dorsal visual stream can become conscious

Conclusions
- Newer work has demonstrated that the original hypotheses of Milner and Goodale were not sufficiently complex
  - For example:
    - Ventral may support online action to familiar objects
    - Dorsal is sensitive to pattern and object information
    - Dorsal lesions can disrupt perception
- Dominant view today is concetualized in terms of distributed networks
Afterthought: Comments from Mel Goodale
- He had a long intellectual journey.
- He worked with non-human animals and humans
- Their original theory has become more complex
- Evolution has led to two visual systems:
  - One indirect system perceptual system linked to action via cognition
  - One direct visuomotor system linked action via motor control processes

Back to Main Advanced Cognitive Psychology