Module 2 Flashcards
Perception
How the external, physical world gets represented in the brain
Agnosia
Inability to recognize objects visually; not a language deficit; does not affect prior memory about objects. Existence of different agnosias suggests that different visual processing uses different brain areas.
Apperceptive Agnosia
Inability to name, match, or discriminate visual objects. Cannot combine basic bits of info into complete images (cannot read)
Associative agnosia
Inability to associate visual patterns with meaning. Can describe, copy, and use items but cannot name the object.
Experience Error
FALSE assumption that what you see is all that exists in the world (ex: illusions)
Steps of Perception
Sensation/input of stimulus, assembling basic visual components, associating meaning
Fixation-saccade cycle
Example of experience error; we use lots of quick jerky eye movements to scan a scene when not tracking an object with smooth pursuit, but we perceive it as continuous.
Bottom-up processing
Data driven; see patterns by building up from basic perceptual input
Top-down processing
Conceptually driven; input is influenced by prior knowledge and experience from memory; allows faster processing. Based on expectations about the world and context of surrounding stimuli
Computational Approach to Perception
Bottom-up theories; perceiving items based purely on sensational input
Template matching theory
We match input with stored templates of items in memory in order to recognize them.
Pros: works to program computer vision, but only if you can control the input.
Cons: Too many templates would need to exist for human perception to function
Feature matching theory
We analyze and match each distinct feature of a visual item to distinguish a whole item, supported by the existence of feature detection neurons
Pandemonium model (1959)
Layers of little ‘demons’ report each feature of an image. Image demon (receives input) -> feature/signal demon (detect and report features) -> cognitive demons (combine features and ‘yell’ if they share many features) -> decision demon (picks the loudest cognitive demon as the perceived image/item)
Pros: Breaking apart and recombining features, serial processing, does not rely on exact matches, parallel processing among ‘demons’
Cons: no explanation on how the features are recombined
Recognition by Components theory (Biederman)
Expansion of matching theory to 3D objects. All objects are made of up 3D shapes called geons. Made to program computer vision for 3D shapes and viewpoint invariance (recognition from any angle of an object)
Non-canonical viewpoints
Viewpoints that are unusual where geons are not perceptible. Makes it difficult or impossible to define an object
View-based recognition
Humans are faster at identifying objects from canonical viewpoints (viewpoint centered). We have neurons specified to certain viewpoints
Cons of Bottom-up Processing
Take too long to process some tasks, can’t explain category discrimination, only covers machine vision
Gestalt Approach to Perception
“The whole is different than the sum of its parts”; how we combine features into objects or events. Heavily observation based, mostly bottom-up
Gestalt grouping principles
Proximity (grouped by distance)
Similarity (grouped by shared features)
Common region (grouped by shared space)
Experience (prior knowledge can affect grouping of certain images forever)
Perception/action approach to Perception ( adapted from Gibson)
Embodied approach, opposes computational approach (argues computers are too far removed from the real world). Argues goal of perception is determining action in the world, not items, and all you need to figure out the world is motion perception. All info comes from the ambient optic array, and objects are recognized by their affordances
Ambient optic array
Structure the world gives to light before it hits the eye (reflections off of objects). We can perceive motion by comparing changes over time in the ambient optic array.
Optic flow
Changes in the ambient optic array (motion)
Distal Stimuli
Anything physically located in the real world
Proximal Stimuli
Mental representations of physical objects
Affordance
What we can do with an object. Affordances vary based on the perceiver of the object
Modern Adaptation of Perception/Action Approach
Both action and representations are needed for perception; action influences how we perceive the world. Perception is embodied, but we don’t know how.
Gibson’s Direct Perception Approach
Affordances of objects directly connect action and perception without need of cognitive processes; representations and memory do not play a role
Ventral Pathway (What)
Object recognition; temporal lobe. Contains specific processing areas (FFA, PPA, etc.)
Fusiform Face Area
Cortical area specified for human faces; right hemisphere
Parahippocampal Place Area
Cortical area specified for recognizing places
Dorsal Pathway (Where/how/action)
Knowing procedural knowledge for objects
Ideomotor Apracia
Deficit in action towards objects/demonstrating how to use objects (can still recognize the object visually)
Blindsight
Lack of conscious vision caused by cortical damage, not eye damage. Can still perform actions towards objects despite not ‘seeing’ them.
Face Processing
We process configurally (as a whole rather than as features), unlike object perception. Facial recognition performance on mutated faces will always be poorer than on an unchanged face. Unknown cause.
Face Inversion effect
We are faster at recognizing upright faces than upside down faces.
Innate view of face perception
We are born with a special face perception system
Experience view of face perception
We have become experts at recognizing faces because we do it so often and have since birth
Diamond and Carey Expert Experiment (1986)
Had dog experts identify upright and upside down images of dog species. Testing if expertise acts the same as face processing. Found that experts did show an inversion effect like face processing. Shoddy replication.
Gauthier and Tarr FFA theory
FFA is evolved to discriminate between highly similar stimuli. Tested with the Greeble experiments (made up shape ‘people’ and testing discrimination of features). Concluded that FFA is for ‘expertise’
