Introduction to Visual Perception

Introduction to Perception

• Perception involves the detection of sensory information and making decisions based upon it.
• It is a complex process with many different, sometimes contradictory, theories.
• These varying theories address different aspects of perception.
• Perception is the subjective experience of sensory information after cognitive processing (attention etc).

Visual Perception

• Light reflected from objects reaches the eye, allowing extraction of information for interaction.
• Light focuses on the retina, at the back of the eye.
• The cornea does most of the focusing instead of the lens.
• The retina is responsible for capturing light and initiating nervous stimuli.
• The human eye is optically simple, with two optical elements
• It is a remarkably complex imaging system
• After the initial processes the data has some impressive post-processing
• Images are inverted on the retina, but are corrected by post-processing.
• Two retinal imaging systems exist: one for low light (night vision) and one for high intensity light (daylight vision).
• The differences in intensity vision are to fit with our environment where there is day and night.
• Rods are photoreceptors for low-light vision and cones for daylight vision.
• Rods are sensitive and non-functioning in bright light. Cones maximally detect different wavelengths of light labeled red/green/blue.
• Cones are feed into opponent color channels, signaling blue/yellow or red/green.
• At low light levels rods are the only device operating and cannot discriminate colour.
• The Purkinje effect is the shift from color discrimination to non-discrimination in low light.
• Visual perception involves reception, transduction, and coding.
• Transduction is the conversion of physical energy into electrochemical patterns in the neurons.
• Coding shows a one to one correlation between the physical stimulis and nervous system activity.

Color Vision and Environment

• Color vision involves physical wavelengths, and point to single wavelength for each wavelength.
• The visual system has evolved to maximize the light properties to interpret environmental information.
• The environment's properties constrains and shapes the visual system.

Gibson's View: Perception for Action

• Argues the function, not the operation, of perception is important.
• The the ecological approach holds perception is how we interact with the world
• The “perception for action” message states sensory information is rich enough to allow interaction without top-down processing.
• Direct perception is perception based only on input, no top-down processing needed.
• Invariants structural are consistent lights and shade.
• Transformational invariant states even as we move the image will "flow" past us.
• Perception stems from moving and interacting within an environment.
• "Perceiving is an action, not a response".
• Object interaction relies on affordances.
• The perceptual process acts largely according to Gibson's manner.
• Affordances are coded automatically.
• Direct perception offers sensible visually guided behavior.
• The visual system is far more than just a "gibsonian" system.

Optic Flow

• Moving toward something causes everything surrounding it to flow past the viewer.
• Helps us judge how fast we are moving as well as avoid obstacles.
• Drivers use changes in tau to control braking
• Tau is the rate of change as an obstacle gets bigger when getting closer.

The Gestalt Approach

• The Gestalt approach emphasizes how objects are defined.
• Figure-ground segregation is key in its usage, recognizing objects mean telling them apart from their surroundings.
• Some patterns of light falling on the retina can be segregated into figure and ground in differing ways.
• The effect of the same stimulus seen in multiple forms illustrates the influence.
• Principles such as proximity are used by laws of perceptual grouping to resolve different objects
• Gestalt approach is limited in how it covers the process of vision, specifically identifying "bits" in the first place.

Feature-Extraction Theories

• Feature detectors describes how incoming stimuli gets detected.
• The Pandemonium model by Selfridge describes a sample outline for the extraction process.
• Pandemonium says there are numerous "demons" detecting features.
• NO demons exist, but cells effectively act as Selfridge's demons to code and transfer data.
• Hubel and Wiesel found some brain cells (visual cortex) react to edges and lines.
• Brain -imaging shows there exists detectors in humans (Haynes & Rees, 2005).
• The code then has to be processed properly to form the object.

Computational Theory Models

• Marr's computational theory suggests object recognition occurs in separate representations.
• 2-D: observer-centered of light-intensity changes (edges contours blobs)
• 2.5-D: observer centered, incorporating gestalt principles, depth
• 3-D: shape recognition that does not use viewpoint.

Biederman's Recognition-by-Components

• It is a model that uses 3- now called geons, to interpret vast ranges of objects.
• A basic aspect is that an object gets recognized once the geons making iy up are ID'd but this.

Parallel Distributed Processing

• PDP accounts for the use of vast combinations of features .
• Patterns can be recognized via activated cells forming a network.
• Any one cell can form parts of numerous networks.
• A brain region, then, may participate in assorted processes.

Specialized Brain Areas

• There are specialized brain areas, such as the fusiform face area.
• Areas show signs of responding to places and bodies
• Fusiform face areas also response (but much less) to other forms.
• A more important process, object recognition has the ability to be adaptive from learning to recognize new objects.

Top Down Influences

• Illusions are a case to observe how various areas break down at various levels.
• Studying illusions provide insight into how we see and interpret outside stimulus with stored knowledge.

Visual Illusions

• It distorts or transforms our view of the world.
• Illusions are classified by Richard Gregory according to bottom- down qualities: those having physical elements.
• "Lowest" illusions - arise during physical phenomena.
• . Mid - level illusions arise from perceptual processes, rather than cognizance, and
• Highest results by knowledge over perception.

Visual Organization

• Muller-Lyer deals with physical perception and Gestalt.
• Illusions in general occur during circumstances where outside information is cut off (Hollow Mask).

Size Constancy & Illusion

• The ability to interpret images while knowing the scaling, distance, etc., of source objects.
• We determine size based to apparent distance; but without that proper scaling.

Knowledge Impact

• Knowledge is determined to have an incredible effect in creating the overall image (Segall et al, 1963).

Honi Phenomenon

• An effect in the brain where those with a long or important partner create the false relative distances that are observed.

LBFS Accidents

• Looked but Failed to See (Sabey & Staughton, 1975; Brown, 2005) are accidents where there may be ample sensory information.
• Thresholds may not be sensed because the stimuli does not line up with awareness.
• There must otherwise be contrast with background.

Sensory & Attentional Interference

• If objects themselves are not readily able to be registered by sensation as an effect of colour and luminance, there becomes a conspicuity.

Attention & Knowledge

• Aspects of the observer play a role to decide whether to consciously interpret the signals.

The Power of Influence

• Past knowledge can influence whether stimuli gets percieved.

Top-Down Processing

• Perceptual hypotheses are examined and then tested with sensory information.

The Visual System

• The two streams are ventral and dorsal.
• Ventral goes to the inferotemporal cortex (what).
• Dorsal goes toward the parietal cortex (where).
• The former helps with recognition , and the latter with spatial info.
• Vision requires both forms.
• Visual interpretations may happen with use of all knowledge.