Theories of Perception

Theories of Perception
Bottom-Up Approaches (Direct)
Top-Down Approaches (Constructive)
What is perception?
• Is it too simple to say we associating what is
seen with what is remembered?
– This is direct perception
• Four types of bottom-up perception
1.Template theories
2.Prototype theories
3.Feature theories
4.Structural-description theories
What is direct perception?
James Gibson
• Sensory information (contextual) is all we need to
perceive
– No higher cognitive processes
• Existing beliefs
• Inferential thoughts
• intelligence
• Emphasized ecological validity
– Labs lack contextual information
• How do we know?
– Infant research
• They develop perception without higher level thinking
Lego Racer Pro/ENGINEER Parts is a good example of bottom-up design
because the parts are first created and then assembled without regard to
how the parts will work in the assembly
Add Context: Top-Down
“Rat Man”
Template Theories
• Definition: highly detailed models for patterns
we potentially recognize
– Stimuli  find template  Perception
• NOT an exact match
• Problems: TOO Rigid
– Our templates need to be adaptable
• We recognize the letter “A” despite its size
– How can we have templates for everything
Prototype Theories
• Definition: Not specific or rigid model
• Strengths
– We can form prototype even though we have not
seen an exact example
• Solso & McCarthy, 1981 (Experiment)
1. Showed them patterns (not prototype)
2. Additional patterns + distracters
3. Identify prototype + confidence in answers
– Implications: Never saw the exact prototype but
were able to identify it because of previous
patterns
Feature Theories
• Local vs. Global perception
• Supported by Biology
• Hubel & Wiesel
– Brain mapping
– Specific neurons for visual field
• Hierarchy: Simple and complex cells
Top-Down Processing
• Constructive
– Conceptually driven
• Definition
– Perceiver builds (constructs) a cognitive
understanding (perception) of a stimulus
– Use sensory information + other info
• Direct ONLY uses sensory information
• AKA  Intelligence perception
– Higher-order thinking is necessary
An example of top down processing: Even though the second letter in each word
is ambiguous, top down processing allows for easy disambiguation based on the
context
Examples
• What do you do if you see a stop sign
that is missing a letter
– Construct  respond by stopping
• Apply to Gestalt Principles
– Size
– Shape
– Color
How do we form and test a hypothesis
1. Sensation (sensory data)
– Prior expectations (expecting to see a friend we
planned to meet)
2. What we know (knowledge in memory)
– What we know about context (trains are on
tracks, cars are not
3. What we can infer (high-level cognitive
processes)
– Based on data and what we know about data
• Correct attributions based on visual
information
– Unconscious inferences
• Unconsciously assimilate info from many sources
– Make judgments were are not aware we are
making
• Correct perception requires….
– Intelligence
– Thought
• Combine sensory info and knowledge from
previous experience
Response to Direct Perception
• It doesn’t fully explain context effects
– Influence of surrounding environment on
perception
• Context effects can be experimented on (lab)
Top-Down Research
• Palmer, 1975
• Method: Experiment
– Control: appropriate context
• Kitchen  bread
– Exp: inappropriate context
• Kitchen  drum, mailbox
• Results
– Appropriate context recognized more rapidly
Configural-Superiority Effect
• improvement in discrimination performance resulting from
the addition of informationless context
stimuli A and B is aided by the addition of context C
Object-superiority effect
• identified a briefly flashed line segment more
accurately when it was part of a drawing that
looked unitary and 3-D
– Compared to if the line was in one of several less
coherent flat designs.
Too Extreme
• Bottom-up
– Does not allow for past experience or knowledge
– Why do we store knowledge?
• Top-down
– Underestimates the importance of sensory
information
• Gross inaccuracies in perception