Theories of Visual Perception: Gibson and Constructivism

Fundamentals of Perception

Definition of Perception: According to Roth ( $1986$ , $p.81$ ), perception refers to the means by which information acquired from the environment via the sense organs is transformed into experiences of objects, events, sounds, tastes, etc.
The Perceptual Path: The process involves the transition from a Distal stimulus (the object in the world) to a Proximal stimulus (the representation on sensory receptors) to the final Perceptual experience.
Importance of Perception:
- It is the only source of information about the physical world.
- All other cognitive systems (memory, reasoning, decision making) rely on the foundational data provided by perception.

The Mechanics of Sensation and Qualia

The Sensation Process: The 'Real World' contains inputs such as Light, Sound, Vibration, Heat, Pressure, Movement, and Chemicals. These are received by sense organs (Eyes, ears, nose, skin, etc.).
Transduction: This is the process where sensory organs convert physical energy into electrical signals for the brain.
Perception (Qualia): The internal subjective experience resulting from transduction.
Visual Encoding Anomalies:
- Light is encoded as an electrical signal starting from the cones in the retina.
- Light itself is not inherently colored; objects are not colored; and the visual signal traveling to the brain is not colored.
- Examples provided for further exploration include Benham's Top (http://www.michaelbach.de/ot/col_benham/index.html) and specific visual phenomena (http://www.youtube.com/watch?v=9_0Z4Bi3JQo).
Is Your Red the Same as My Red?:
- Explanatory Gap: The difficulty in explaining how physical processes in the brain give rise to subjective conscious experiences.
- Pain Example: Subjectivity in how individuals experience physical discomfort.
- Synesthesia/Tastes: Considerations of tasting colors or differences in how individuals perceive the same taste.

Comparative Perception in Other Animals

Sensitivity Variations: Animals often have higher or lower sensitivity compared to humans.
Canine Perception: Dogs have $50 \times$ more scent receptors than humans, and approximately $40\%$ of their brain is dedicated to processing smell.
Different Sensory Modalities: Some animals use senses humans do not possess:
- Electric and Magnetic field detection.
- Different frequency ranges: Bats perceive ultrasonic frequencies, while bees perceive ultraviolet light.

Theoretical Approaches to Visual Perception

The Core Question: How is visual perception achieved from retinal information?
Major Theories:
- The Gestalt approach: Focuses on laws of organization (e.g., the whole is more than the sum of parts).
- Gibson’s ecological theory: Emphasizes direct perception and environmental interaction.
- Marr’s information processing theory: A computational, bottom-up approach.
- The Constructivist approach: Emphasizes top-down processing and stored knowledge.
Differentiators: These theories vary based on their reliance on Bottom-up vs. Top-down processing, the defined goal of perception (e.g., action vs. reconstruction), and the methods of study (e.g., lab-based vs. natural environment).

Gibson’s Ecological Theory of Perception

Core Tenets:
- Bottom-up approach: Perception is 'direct' and does not require complex cognitive processes.
- Sufficient Information: The richness of the retinal array provided by the environment is enough for successful interaction.
- Direct Perception: Perception is viewed as identical to sensation; the environment 'affords' perception without mediation.
- Ecological Method: Perception should be studied in natural environments rather than laboratories using $2\text{D}$ displays. Movement of the observer must be permitted.
Historical Influences:
- William James ( $1842-1910$ ).
- Edwin Holt ( $1873-1946$ ).
- Kurt Koffka ( $1886-1941$ ) of the Gestalt School.
- Eleanor Jack Gibson ( $1910-2002$ ), known for the visual cliff experiments (married to James Gibson).
Evolution of Ideas:
- James Gibson ( $1904-1979$ ) was inspired by train rides as a child, noticing how the world 'flowed' and 'expanded' toward him while 'contracting' behind him.
- He applied these observations of Optic Flow to pilot training for the military.
- Emphasized that the active movement of the observer provides a continuous stream of information.
Ambient Optic Array: The structure of light reflected by textured surfaces in the environment, which changes as the observer moves.

Invariants and Information in the Environment

Invariants: Unambiguous information about the environment that remains constant and can be directly perceived.
Example 1: Horizon Ratio Relation:
- The proportion of an object above and below the horizon line remains constant for objects of the same size standing on the same ground, regardless of distance from the observer.
Example 2: Texture Gradients:
- Changes in texture in the optic array provide information about distance, orientation, and surface curvature.
- Texture becomes smaller and finer as distance increases.

Empirical Evidence: Gibson & Bridgeman ( $1987$ )

Study Design: Participants were asked to identify properties of objects based on black-and-white photographs of surfaces.
Questionnaire and Accuracy Rates:
- 1. Substance identification: $67.0\%$ correct ( $\\chi^2 = 201.9$ , P < 0.01).
- 2. State of matter (Gas/Liquid/Solid): $84.0\%$ correct ( $\\chi^2 = 1.01$ , P > 0.10).
- 3. Deformability (Bend/change surface): $49.8\%$ correct ( $\\chi^2 = 2.01$ , P > 0.10).
- 4. Surface planarity (Flat/Uneven): $83.1\%$ correct ( $\\chi^2 = 191.1$ , P < 0.01).
- 5. Pigmentation consistency: $72.1\%$ correct ( $\\chi^2 = 85.33$ , P < 0.01).
- 6. Lighting type (Diffuse/Directional): $59.0\%$ correct ( $\\chi^2 = 14.28$ , P < 0.01).
Results: The average subject identified about two-thirds of the photographs correctly, supporting the idea that static surface information (texture) is highly informative.

The Role of Motion in Gibson’s Theory

Importance of Motion: Motion is necessary to highlight invariant properties. In a static scene, everything is invariant, thus motion differentiates properties.
Observer vs. Object Motion: Gibson focused on observer movement because it generates significant changes in the ambient optic array.
Motion Parallax:
- A monocular cue to depth.
- As an observer moves, nearby objects appear to move rapidly, whereas far objects appear to move slowly.
- This cue is used extensively by animals with limited binocular overlap (e.g., birds and lizards) through 'head bobbing' or 'orthogonal running'.
Optic Flow (Expansion):
- A combination of parallax and changes in retinal size.
- Used in modern technology like drones and autonomous self-driving cars to navigate and avoid collisions.
Practical Application: Horizontal lines are painted on roads near junctions or motorway exits. As they become closer together, they create the illusion of increasing speed (optic flow), prompting drivers to slow down.

Affordances: Perception and Action

Definition: An affordance is an opportunity for action; it is what a surface or object offers the animal (e.g., graspable, edible, sit-on-able).
Nature of Affordances: They are actionable possibilities relative to the organism. A chair is not perceived as a 'chair' abstractly, but as 'sit-on-able'.
Action-Specific Perception Evidence:
- Warren ( $1984$ ): Perception of whether a doorway is passable depends on the ratio of aperture width to shoulder width ( $\\sim 1.3 \times$ threshold). Whether a step is climbable depends on the ratio of riser height to leg length (critical ratio of $0.88$ ).
- Proffitt et al. ( $2003$ ): Hills appear steeper to observers when they are tired or carrying heavy loads (encumbered).
Neurological Evidence:
- Mirror Neurons: Fire when observing others perform actions.
- Canonical Neurons: Respond to objects in terms of potential actions (size, orientation, grasping).
- Peripersonal Space: The brain recalibrates the perception of space near the body when using tools.
Key Problem for Affordances: Gibson argues affordances are direct, but some objects have multiple affordances (e.g., a chair can be used to sit, stand, throw, or block a door). Choosing one over the other suggests top-down processing involving goals and context.

Gibson’s Theory: Summary and Criticism

Summary:
- Highlighted depth and distance information in the retinal image.
- Claimed perception is bottom-up and 'direct'.
- Emphasized action as the primary goal of perception.
Criticisms:
- The mechanisms for how information is 'picked up' are vague.
- Ignores the role of top-down processing (experience and memory).
- Largely ignores neuroscience.

The Constructivist Approach to Perception

Core Tenets:
- Retinal images are impoverished and do not provide sufficient information.
- Perception depends on stored knowledge, memory, and experience.
- It is a Top-down and Bottom-up hybrid.
Hermann von Helmholtz ( $1821-1894$ ):
- Unconscious Inference: An involuntary, pre-rational, reflex-like mechanism that forms visual impressions.
- Likelihood Principle: We perceive the object that is most likely to have caused the retinal pattern.
- Fixed Neural Processing: Evidenced by illusions (like the Titchner Illusion where brain over-emphasizes size differences) that are impervious to experience.
Richard Gregory ( $1923-2010$ ):
- Described Helmholtz as the modern founder of perception science.
- Perceptual Hypotheses: The brain generates hypotheses about what it sees. Illusions (like the Necker Cube or Hollow Mask) occur when stored knowledge leads to inaccurate hypotheses.
- Hollow Mask Illusion: Explained by our prior knowledge that faces are always convex; we cannot see it as hollow even if it is.
Critique: Like Gibson's theory, the constructivist approach is criticized for being vague regarding how hypotheses are generated.

Comparative Summary of Perceptual Theories

Feature	Gestalt	Gibson	Marr	Constructivist
Processing	Bottom-up & Top-down	Bottom-up	Bottom-up	Bottom-up & Top-down
Goal	Reconstruction	Action	Reconstruction	Reconstruction
Methods	Experimental/Lab	Outside Lab/Experimental	Computational	Experimental/Lab
Key Aspect	Sum more than parts	Direct Perception for action	Computational/Information problem	Perceptual Hypotheses
Criticism	Descriptive, not explanatory	Vague pickup mechanisms; no top-down	Vauge; missing top-down role?	Vague; how are hypotheses generated?

Key Terminology Recap

Bottom-up processing: Data-driven, starting from the sensory input.
Top-down processing: Concept-driven, relying on knowledge and expectations.
Ambient Optic Array: Gibson's term for the structure of light in the environment reaching the eye.
Invariants: Environmental properties that remain the same across different viewpoints.
Affordance: Actionable possibilities offered by the environment.
Marr's Descriptions: Grey-level description, Primal sketch, $2 \frac{1}{2}\text{D}$ sketch, and $3\text{D}$ object-centered description.
Perceptual Hypotheses: (Gregory) The brain's guesses about the contents of the visual world.

References and Suggested Reading

Bruce, Green & Georgeson ( $2010$ ): Visual perception: physiology, psychology and ecology, Chapters $4$ & $10$ .
Gibson, J.J. ( $1979$ ): The ecological approach to visual perception.
Gregory, R.L. ( $1980$ ): 'Perceptions as hypotheses', Philosophical Transactions of the Royal Society of London.
Marr, D. ( $1982$ ): Vision.
Palmer ( $1999$ ): Vision Science: photons to phenomenology, Chapter $2$ .
Snowden, Thompson, & Troscianko ( $2006$ ): Basic vision: An introduction to visual perception.

Which of the following best describes Gibson's ecological theory of perception?
- A) Perception relies heavily on cognitive processes.
- B) Perception is directly linked to sensory input and environmental interaction.
- C) Perception is primarily a mental construct based on past experiences.
- D) Perception involves complex cognitive inference.
What does the term 'affordance' refer to in Gibson's theory?
- A) The physical characteristics of an object.
- B) The opportunities for action that an object or environment provides.
- C) A direct perception without the need for cognitive processing.
- D) The subjective interpretation of sensory data.
According to the Constructivist approach, how do we perceive the world?
- A) Through direct sensory experiences that require no cognitive input.
- B) By relying solely on invariant properties of stimuli.
- C) By combining sensory input with stored knowledge and experiences.
- D) By focusing exclusively on the environment's physical qualities.
Who is credited with the concept of 'unconscious inference' in perception?
- A) James Gibson
- B) Richard Gregory
- C) Hermann von Helmholtz
- D) Eleanor Jack Gibson
Which statement reflects a critique of Gibson's theory?
- A) It emphasizes the role of top-down processing in perception.
- B) It successfully explains how affordances function in everyday life.
- C) It provides a clear explanation of how perceptual mechanisms work.
- D) It largely ignores neuroscientific evidence related to perception.
The Constructivist approach is considered a hybrid of which types of processing?
- A) Only bottom-up processing.
- B) Only top-down processing.
- C) Both bottom-up and top-down processing.
- D) Neither; it relies on behavioral responses.

Answers:

Which of the following is a key principle of Gibson's Ecological Theory of Perception?

A) Perception is formed through learned experiences.

B) Perception occurs directly from environmental interactions without cognitive mediation.

C) Perception is largely a result of memory and prior knowledge.

D) Perception is influenced by social and cultural factors.

What do "invariants" refer to in Gibson's theory?

A) Elements that change with perspective.

B) Constant features of the environment that can be directly perceived.

C) Subjective interpretations of sensory experiences.

D) Variations in sensory modalities across different species.

Which of the following best characterizes the Constructivist approach to perception?

A) Perception is solely determined by immediate sensory data.

B) Perception is driven by high-level cognitive processes and context.

C) Perception is mainly a bottom-up process without any top-down influence.

D) Perception is a primary instinctual response, unaffected by past experiences.

According to Richard Gregory, what is a significant aspect of perception?

A) It is purely a physiological process.

B) It is based on hypotheses generated by the brain.

C) It is entirely dependent on the immediate environment.

D) It requires no prior knowledge or learning.

In the Constructivist framework, how do perceptual hypotheses affect visual perception?

A) They simplify the sensory input for immediate understanding.

B) They can lead to misinterpretations and optical illusions.

C) They are irrelevant to the perception of objects.

D) They provide direct, unverifiable information about the environment.

Answers: