PSY 420 Exam 4

Sensation and Perception Final Exam

Inverse projection problem

Retinal image is ambiguous → an object can create an infinite number of responses

Edge at conceptual level

area in visual field that is high contrast, a clear line separating areas

How can the ambiguity of the retinal image (sensations) be resolved?

by two ways → bottom-up and top-down processing

need more information → constraints and assumptions

Buttom-up processing

information flows UP from lower levels to higher levels of analysis, integrating simple sensory attributes into large structures on the basis of built-in rules

rules are hard-wired into nervous system (ex. Gestalt grouping rules)

Gestalt grouping rules

set of rules describing when elements in an image will appear to group together by looking for patterns (contours, edges, what’s available)

rules of thumb → apply most of the time, but there are exceptions

ex. law of similarity, law of good continuation, law of proximity (nearness), law of familiarity

Law of similarity

similar things (ex. color, shape, distance, size) will be grouped together

Law of good continuation

smoothest path determines sameness

Law of proximity (nearness)

closely spaced things are grouped together

Law of familiarity

if the collection of parts is meaningful and familiar, it forms a group

Top-down processing

information flows DOWN from higher levels to lower levels of analysis, using prior knowledge and experience to steer lower level processes

perception fits incoming info. rather than having to deduce it

we know when things are physically impossible, so brain doesn’t consider any other option (ex. Light-from-above heuristic, size difference in pennies, and Eiffel tower)

Light-from-above heuristic

assumptions that light sources are typically above our head (sunlight, light fixtures) → implacates in how we interpret shadows, and shapes based on those shadows

assumption can be exploited in illusions

Eiffel tower top-down processing

A giant hand isn’t picking up the Eiffel tower because we know exact size of the structure

interpretations depend on background knowledge

To study higher level perceptual processes deaing with ambiguity

have to use stimuli carefully designed to create specific kinds of ambiguity

ex. bistable, rivalrous, and impoverished stimuli

Bistable stimuli

stimuli presenting two equally plausible interpretations

visual ex. → necker cube (is the green part in front or back)

auditory ex. → auditory streaming (two streams when fast and one stream when slow) and verbal transformation (repeating same word turns into a different word)

Bottom/Top processing evidence of bistable stimuli

bottom-up explanation → both stimuli compete, one becomes dominate, adaptation takes over when dominant is fatigued, and perception keeps alternating

top-down explanation → indecision in higher-level processes and interpretations switch depending on attention

Studies showing evidence of processing of bistable stimuli

auditory streaming → bottom-up processing; adaptation in frequency-selective cortical neurons

verbal transformation → top-down processing; transitions linked to frontal lobe (attention)

visual stimuli → both bottom-up and top-down processing

Rivalrous stimuli

two unrelated stimuli are presented at the same time to different sensory receptors, which would never naturally be present together

visual ex. → two different images presented to two eyes

smell ex. → two different odor compounds presented to two nostrils

hearing ex. → two tones of different pitch are presented to two ears

Processing evidence for rivalrous stimuli

one image, tone, odor dominates for a while and then is replaced by the other one

evidence for both adaptation (bottom-up) and top-down (attention) processing for alternating perceptions

Impoverished stimuli

stimuli with so little information that it may be difficult or impossible to interpret them without additional knowledge

used to explore role of top-down information flow when resolving ambiguity → interpretation correlated with activity in secondary visual cortex

once the “blobs” have been identified as a face, the image remains “organized” in subsequent presentations

Evidence of top-down influence in impoverished stimuli

interpretation of full picture (higher level process in secondary visual cortex) affects activity of low level processes in primary visual cortex

involves both feedforward signals from low-high levels of processing and also feedback from high level areas down to lower levels

Unconscious inference

Helmholtz’s explanation of how we use constraints/knowledge → logical process of drawing conclusions by combining incoming sensory data with memories of past experiences that match them

sensations are incomplete or incorrect, so → the matching process fixes these problems and provides an accurate representation of the real world

at the heart → is the application of Bayes rule: the combination of prior probabilities and likelihoods

Bayes’ Theorem of resolving ambiguities in sensory processing

inductive logic → a general conclusion is drawn from a specific set of facts; similar to hypothesis-testing

ex. '“swans i’ve seen are white, therefore all swans must be white”

important terms → prior and likelihood

Prior

probability of a particular perceptual interception based on past experience (swans are white)

Likelihood

probability of a particular interpretation based on current sensory information (seeing a white swan)

Bayesian inference in perception

goals → calculate most reliable estimate of an object in the face of uncertain sensory information

where → sensory data supplies likelihood and built-in knowledge supplies prior probability

likelihood + prior = posterior probability

Posterior probability

probability that a certain interpretation is correct based on combining priors and likelihood for interpretation

Maximum posterior probability

interpretation with the highest probability

Role of feedback in the Bayesian approach

involves using prior knowledge to update interpretations based on new sensory information to improve the reliability of perceptual judgments

perception is a trade-off between prior knowledge and the reliability of the incoming data

P(SII) = P(S) x P(IIS) (posterior = prior x likelihood)

Does the brain use Bayesian computation?

basic assumption → brain represents perceptual information probabilistically

population codes store relative probabilities of different stimulus values

population code for likelihoods + another for priors

Coding efficiency

minimization of ambiguity maximizes coding efficiency

top-down signals promote → once high-level representation is activated, the low-level activity becomes redundant

Perceptual constances

movement changes retinal images but brain still sees stable 3D objects due to inferences

Univariance

a problem that occurs when scientists measure neural activity from a single neuron → neuron responds to a lot of different stimuli, so there will be some cases where the neuron will produce the same amount of neural impulses to two different stimuli

not a problem for the perceiver → the perceiver uses groups of neurons (a population code) to respond to uniquely to a given stimulus

The perceptual process

stimulus generates a retinal image, which has to be matched to real object’s stored representation in cortex (assuming incoming info. is incomplete/incorrect and brain has to interpret from that)

analogy → “unscrambling scrambled eggs” → brain has to unscramble the scrambled retinal image

Top-down explanation of motion ambiguity

attention (a high-level cognitive process) determines which interpretation will dominate at any one time → can be driven by eye movements, as they attend to different parts of the stimulus

neuroimaging evidence supports this because during the switch from one interpretation to another, the parietal lobe is highly active → which controls visual attention

it’s a top-down influence → a high-level process overrides and reinterprets the raw stimulus.

7 questions of ecological psychology

1. Who perceives?

2. Where do we perceive?

3. What are the units of perception?

4. How do we perceive?

5. What is perception for?

6. What variables are perceived?

7. Where does richness of perception come from?

1. Who perceives?

the WHOLE organism, not just a sense organ

2. Where do we perceive?

perceiving takes place in an environment, not just in the head

“animal-environment interaction”

3. What are the units of perception?

the entire organism-environment system, not just neural impulses, receptor cells, sense organs, and/or brain

4. How do we perceive?

by active exploration by perceptual systems, NOT by passive stimulation of sense organs

eyes, head, neck, torso, and legs are ALL part of the visual system because all can affect what is seen by the perceiver

visual cliff exploration

5. What is perception for?

perception is for the control of action, NOT for the conscious seeing, hearing, feelings, smelling, and tasting of objects

concept of affordance

6. What variables are perceived?

invariants → complex (higher order) patterns of physical stimulation

we perceive affordances specified by invariants, NOT simple physical variables like amplitude/intensity, length, etc.

variables of relevance are global, wholistic patterns in the environment - proximal status is ignored

1:1 mapping

7. Where does richness of perception come from?

it comes from the richness of the stimulation, NOT from the complexity of the perceptual processing itself

if stimulation is rich and unambiguous, it needs no further processing

Organism-environment system

organism and environment are inseparable and form only one unitary system → mental activity is activity of the whole system

includes → receptors, sense organs, brain, and much more (such as structured environment)

Visual cliff experiment

older infants avoided the “deep” side, whereas younger infants didn’t

perceiving affordances for a steep drop-off improves with more crawling experience (experience by exploration)

Exploratory Procedures (EPs) by Lederman & Klatzky

linked to specific object properties two ways → EP associated with an object property is (1) executed spontaneously when information about that property is desired and (2) appears to optimize info. uptake about that property

Exploratory Procedure: Lateral motion

associated property → surface texture

behavior → the skin is passed laterally across a surface, producing shear force

Exploratory Procedure: Pressure

associated property → compliance or hardness

behavior → force is exerted on the object against a resisting force; ex. by pressing into the surface, bending the object, or twisting

Exploratory Procedure: Static contact

associated property → apparent temperature

behavior → the skin surface is held in contact with the object surface, without motion; typically a large surface (like the whole hand) is applied; this EP gives rise to heat flow between the skin and the object.

Exploratory Procedure: Unsupported holding

associated property → weight

behavior → the object is held while the hand is not externally supported; typically this EP involves lifting, hefting or wielding the object

Exploratory Procedure: Enclosure

associated property → volume; global shape

behavior → the fingers (or other exploring effector) are molded closely to the object surface

Exploratory Procedure: Contour following

associated property → exact shape

behavior → skin contact follows the gradient of the object's surface or is maintained along edges when they are present

Affordances

of the environment → what it offers animals, what it provides or furnishes, either for good or ill

are opportunities for action for a given organism depending on the an animal’s behavioral repertoire (aka action capabilities)

perceiving them → perceiving what is possible, like telling the future

highlights the animal-environment interaction

Perceiving affordances means

to perceive surfaces in the environment in body-scaled items

information → relevant to behavior, body-scaled, and rich

π-number

riser height (cm)/leg length (cm) = pi (a dimensionless number without units)

can perceive opportunities for action and perceive opportunities for most efficient action

3 ways to describe the link between perceiver and environment

exteroception, proprioception, exproprioception

energy arrays must be rich enough to accommodate these perceptions

Exteroceptive properties

spatial location of objects in environment

ex. is the ball reachable

Proprioceptive properties

self → relative position of limbs

ex. how is my left arm positioned

Exproprioceptive properties

Self-environment → limbs relative to environment

ex. am I choked up enough on the racket

1:1 mapping between stimulation pattern and event

animal-environment relation generates → information, which specifies → animal-environment relation

information X means only event Y has occurred → produces a reduced need for processing, so perception is direct

if you do not find a single-valued relations between a physical variable and a perceptual experience → keep looking

Invariants

aspects of the environment that always stay the same → if info is reliable, then we need to look for aspects that are constant

it is the perceiver’s job to discover and “lock on to” them → change is often required to reveal one (“biological motion”)

horizon-ratio

Horizon-ratio as an invariant for perceived size

horizon ALWAYS appears to intersect at the viewer’s eye height

can help determine size → is bigger if object passes line, is smaller if below line

Invariants are independent of:

1) situational factors, 2) species, and 3) perceptual system

Smart Perceptual Device

capitalizes on invariants (without intermediary processing or calculating)

ex. Polar Planimeter → measures areas of irregularly shaped objects without calculation

Perceptual learning

is the development of perceptual skill with practice

ex. The “Scribble” Experiment → subject improved with practice (even without feedback)

Ambient optic array

the pattern of light reaching the eye from the environment, containing all the visual information about surfaces, textures, and objects in a scene

it's the source of visual data that our brain interprets to understand the environment

at an observation point → set of solid optical angles; shows how the perceiver is situated

Optic flow

the apparent motion of elements in the optic array as an observer moves through the environment

helps in perceiving motion, distance, and direction, such as how objects seem to "move" when walking forward

How does Gibson account for Motion Perception?

perceived motion is a result of local patterns of change in the optic array

some local patterns of change in optical texture specify motion across an observer, while others specify motion toward an observer

Determining direction of motion from optic flow patterns, “rules” for behavior

symmetric expansion → the object is directly approaching you and will hit

• to catch object → arrange self so the optical texture is expanding symmetrically

asymmetric expansion → object is approaching you but will miss

• to avoid object → arrange self so optical texture is expanding asymmetrically

Global change in the optic array

tells about motion of self → certain pattern of expansion + occlusion = locomotion (movement)

transformation of optic array uniquely identified the motion as motion of objects or of me, my eyes, or my head

focus of expansion (FOE) → location is direction you are heading

Rules for the visual control of locomotion include:

1) to start, make the optic array flow by initiating movement; to stop, cancel the flow (no movement); to go back, make the flow reverse

2) to turn, shift the center of outflow from one patch in optic array to another; to steer, keep center of outflow from obstacles and brinks

Perception

is the active selection of information for controlling action

Perception-action cycles

a continuous process in which sensory perception guides motor actions, and those actions change the environment, leading to new sensory input

it highlights the dynamic interaction between perception and behavior to achieve goals