Perception Flashcards
Lens
Focuses light as it enters the eye
Retina
Contains photoreceptors (rods and cones) which transduce light into electrical signals (action potentials)
Fovea
The centre of the retina which contains the highest density of cone receptors. Supplies the brain with fine-
grained details and colour
Visual pathway
Visual information travels from eyes to Lateral Geniculate Nucleus (LGN) -> Calcarine sulcus in V1 -> Feature detectors -> hierarchical processing from V1 to the temporal and parietal lobes could bind line features into complex shapes
Contralateral mapping
Light coming from the left side of space (left visual field) is sent to the right hemisphere (left visual cortex) and vice versa
What makes visual perception hard? Solution?
- Objection constancy
- Inverse projection problem
Solution: Unconscious Inference
Object constancy
The very same object can project a different image on your retina because of:
- Viewing angle
- Lighting
- Distance
- Partial occlusion
Inverse projection problem
Each 2D retinal image could reflect infinitely many distinct 3D stimuli
Likelihood principle
We perceive the object that is most likely to have
caused the pattern of stimuli that we received
- Views perception as a form of problem
solving
- We unconsciously and automatically draw
inferences that are based on experience and
basic principles of simplicity
What does the likelihood principle allow?
Top-down biases on perception (top-down processing = Processing that involves a person’s knowledge or expectations)
Bayesian Inference Models
The idea that our estimate of the probability of an outcome is determined by the prior probability (our initial belief) and the likelihood (the extent to which the available evidence is consistent with the outcome).*
Gestalt psychology
Involves heuristics (a “rule of thumb” that provides a best-guess solution to a problem) used by the visual system in order to perform figure ground segregation (identifying a figure from a background)
Gestalt principles
- Similarity
- Closure
- Proximity
- Good continuation
- Good figure/simplicity
Similarity
Points which share features are grouped together
Closure
Points form closed (whole) objects
Proximity
Points close in space are grouped together
Good continuation
Points are assumed to be connected by smooth lines
Good figure/simplicity
Every stimulus is seen as being as simple as possible
Environmental Regularites
- Physical regularities
- Semantic regularities
Physical regularities
Regularly occurring physical properties of
the environment. Examples are light from above assumption and oblique effect
Light from above assumption
Assume that 3D objects are lit from above
Oblique effect
People are better at processing horizontal and
vertical lines because they are more common than oblique. Caused by neural plasticity
Semantic regularities
The characteristics associated with the
functions carried out indifferent types of scenes. Example is Palmer’s scene schema experiment
Scene schema experiment
Participants recognize schema consistent objects (bread) 80% of time but incongruent objects (mailbox) only 40% of the time
Monocular depth cues
- Familiar Size
- Occlusion and Interposition
- Texture Gradient
- Linear Perspective
- Relative Size/Height
Binocular depth/distance cues
- Convergence: the extent to which your eyes are turned inward to fixate an object. Only works for short distances (<1m)
- Stereopsis/retinal disparity: the difference in image projected to the left and right eye; used to make 3D movies appear 3D
Theories of object recognition
- Objected centred
- Viewer centred
Object centred theories
Propose that we store one 3D template for each object (parts + their relationships)
Viewer centred theories
Propose that we store many view-specific templates for each object (e.g. at different angles)
Pros of object centred theories
- Only requires one representation per object
- Can explain why some views are hard to recognize (hidden geons)
Cons of object centred theories
- Can’t discriminate objects that share geon configurations
- Absence of physiological evidence
Pros of viewer centred theories
- Natural views exist
- Recognition time depends on viewpoint
- Aligns with physiological data
Cons of viewer centred theories
Requires a lot of stored representations
So object centred or viewer centred?
Aspects of both theories seem to be important & the brain could do both. Now researchers try to understand how people learn to infer 3D shapes from multiple viewpoints
What is perception for? (Marr’s computational level)
- Perception evolved so that animals could act on their worlds
- Suggests that perceptual and motor systems must be intimately connected
How do actions help us perceive?
- The inverse projection problem becomes less of a problem when you can move around
- Multiple viewpoints help us to better infer the two 3D structure of objects
Object Discrimination
“What” task - monkeys had to select food wells based on the object covering the well
Landmark Discrimination
“Where” task - monkeys had to select food wells
based on their closeness to the landmark object
Interactions between perception and action streams
- Mirror neurons
- Functional connectivity
Mirror neurons
- di Pelligrino et al discovered that single neurons in the monkey premotor cortex fire when monkeys perform OR perceive a specific action!
- Mirror neurons have since been argued to support empathy and theory of mind: the representation of other’s intentions
- Debatable whether humans have them
Functional connectivity
- Brain regions in the dorsal and ventral stream talk to each other during perception and action
- Neuroimaging studies measuring functional connectivity show that information in these pathways could be integrated through connectivity
Neural plasticity
How changes in neural activity after learning a new stimulus occur
“What” pathway
Striate cortex -> temporal lobe, object discrimination
“Where” pathway
Striate cortex -> parietal lobe, landmark discrimination