Lecture 2 Notes: Theories of Perception

Theories of Perception

Objectives

• Explore different issues in attempts to understand the nature of perception.
• Focus on the difference between passive and active theories of perception.
• Distinguish the “Old Look” and “New Look” schools of perception.
• Distinguish the minimum and likelihood principles.
• Examine the idea of perception as a process of unconscious inference.

Some Distinctions

• Passive versus active theories.
• Old Look versus New Look.
• Bottom-up versus top-down processes.
• Stimulus-driven versus knowledge-driven.
• Perception versus cognition. The question is raised whether perception can be discussed independently of cognition.

Rene Descartes

• Reference to Rene Descartes and his famous quote "I think therefore I am" - all we can really assume is that we’re thinking.
• Alternative view: Naive realism.

Neisser (1967) and Naive Realism

Neisser (1967) defined three beliefs about visual perception under the heading of “naïve realism”:

1. Visual experiences mirror the external stimulus. This is related to hallucinations.
2. Visual experience starts and ends with the onset and offset of the external stimulus, relating to visual persistence and visual memory.
3. Visual experiences are based on passive copies of the outside world which could be described using verbal reports.

Passive, Bottom-up Processes in Perception

• Selfridge’s (1959) – Model of visual letter recognition.
• “The pandemonium model”
• How do we recognise things in vision? A strictly bottom-up account

Passive Perception

• Distal stimulus.
• Proximal stimulus.

Modularity of Mind Hypothesis - Fodor (1983)

• The modularity of mind hypothesis as described by Fodor (1983) involves input modules for touch, taste, smell, sight, and hearing.
• Touch: Input modules $To<em>1$, $To</em>2$.
• Taste: Input modules $Ta<em>1$, $Ta</em>2$.
• Smell: Input modules $Sm<em>1$, $Sm</em>2$.
• Sight: Input modules $Si<em>1$, $Si</em>2$.
• Hearing: Input modules $He<em>1$, $He</em>2$.
• The sensory information undergoes transduction and sensory encoding before reaching central processors leading to thought, interpretation, perception, and cognition.

Gestalt Laws of Perceptual Organisation

• No principle dominates.
• Grouping by proximity.
• Grouping by similarity (colour).
• Grouping by similarity (form).

Gestalt Laws - Bottom-Up Processing

• These laws apply from the bottom-up – without knowledge of what the possible objects in the scene are.

Grouping by Common Region (Palmer, 1992)

• Grouping by Common Region described by Palmer in 1992.

Contradiction to Naive Realism

Naïve realism, specifically the idea that visual experiences are passive copies of the outside world describable by verbal reports, is contradicted by evidence from the Gestalt principles. The sensory information is organized according to certain coding principles that operate to facilitate object recognition. These principles break down the input into potential objects.

Innate Perceptual Organization

• Quinn, Burke & Rush, 1993 studied 3 month old infants
• Gestalt psychologists believed that perceptual organisation was innate, and therefore not due to top-down influences.

Old Look versus New Look

• The work of the Gestalt Psychologists can be referred to as the “Old Look” school of perception.
• Theorists interested in passive coding accounts.
• Bottom-up.
• Stimulus-driven.
• Contrasted with the approach of Bruner and others: “New Look” approach to perception.
• Top-down.
• Knowledge-driven.

Bruner and Postman (1949)

• Cards were briefly presented.
• Normal cards were more accurately recognised than incongruous cards.
• An effect of expectation on perception?

Knowledge / Familiarity

• Present single letters briefly and masked.
• Participants say what they see.
• Common letters are better reported than rare ones.

Knowledge / Familiarity - Novel Cases

Distinction between “seeing” and “seeing as” – the difference between perceptual and interpretative operations. The process of seeing is the same for both familiar and unfamiliar cases, yet something else appears to be going on when we are confronted by familiar entities as compared with unfamiliar entities.

Ambiguous Figures

Example of an ambiguous figure that can be seen as a young woman or an old woman.

Epstein and Rock (1960)

Epstein and Rock (1960) research question: is recency or expectancy the over-riding factor in determining the current perception? The participant was presented with a sequence of stimuli (ie. Old, Young, Old, Young …..) and the sequence ended with the ambiguous version. What do you see?

Recency versus Expectancy

Current interpretations of the ambiguous figure tended to reflect an effect of recency rather than what was expected on the basis of the alternation. The question is whether recency or expectancy is driving perceptions.

Minsky’s (1975) Frame Theory

Minsky’s (1975) Frame theory of perception focuses on the role of expectancy. Based on our knowledge of the world we actively generate expectations about what will occur next. Such ideas are present in both theories of vision and audition.

Bruner’s Perceptual Readiness Theory

• Need and value determine our perceptions of the world.
• Our perceptions of the world reflect how we construe the world …

Bruner’s Perceptual Readiness Theory - Bruner and Goodman (1947)

Bruner and Goodman (1947) conducted an experiment where participants had to alter the size of a spot of light so that it matched a comparison disc. The disc varied in size and nature. Generally good estimates of size when the discs were neutral, but this changed dramatically when the discs were replaced with coins.

Bruner and Goodman (1947) - Coin Size Estimation

Participants generally overestimated the size of coins, and these estimates tended to increase with the value of the coins. Poor children’s estimates were more extreme than the rich children, and their overestimates increased more dramatically with the value of the coins.

Simplicity versus Likelihood?

Do we perceive what is given by the simplest reading or the most likely reading?

Minimum Principle versus Likelihood Principle

• The minimum principle: “We perceive whatever object or scene would most simply or economically fit the sensory pattern” (Hochberg, 1981, p263)
• The likelihood principle: “We perceive whatever object or scene would, under normal circumstances, most likely fit the sensory pattern” (Hochberg, 1981, p263)

Light from Above Assumption

A general assumption is that light tends to come from above, and this determines which aspects of the stimulus are seen as hollows and which are seen as protuberances. The minimum principle fails to account for this because light coming from above is no simpler than light from below, yet the perceptual bias of light shining down determines the interpretation.

Four Lines or Two Triangles?

• Two triangles more likely.
• Operation of Gestalt principles again.

Necker Cube

Example of a Necker cube.

Peterson and Hochberg (1983)

Simplicity versus Likelihood? Peterson and Hochberg (1983) found that the interpretation of the figures depends on where the participants fixate. Therefore, there are factors other than minimal coding at work.

No General Principle of Perceptual Organisation?

Maybe there is no general principle of perceptual organisation.

Crude-to-Fine Distinction

• Global-to-local processing.

Global vs. Local Processing Experiment

Examples of composite letters as used in a typical global-to-local processing experiment. Includes consistent, inconsistent and neutral conditions.

Global Preference

• Reaction times shorter when classifying global than local.
• Typically there is a global-to-local interference.
• Reaction times are slowed when letters “point” to different responses.
• Hardly ever are there cases of local-to-global interference.

Perceptual Output

Illustrates the time course of global and local analysis, with global analysis occurring before local analysis.

Visual System Distinguishes Crude from Fine Levels

How the visual system distinguishes crude from fine levels of analysis (from Watt, 1988).

Two General Accounts of Perception

Two quite different and general accounts:

1. Modularity of mind.
2. Interactive hypothesis generation and testing.

• Includes aspects of sensory input, transduction, central processors, thought, interpretation, perception, cognition, feedback, perceptual knowledge, conceptual knowledge, and consciousness.
• Highlights top-down and bottom-up paths, hypothesis generation, pre-processing, and the exploration of objects.

Objectives Revisited

• Explore different issues in attempts to understand the nature of perception.
• Focus on the difference between passive and active theories of perception.
• Distinguish the “Old Look” and “New Look” schools of perception.
• Distinguish the minimum and likelihood principles.
• Examine the idea of perception as a process of unconscious inference. Can we talk about perception in the absence of cognition?