UConn PSYC 3501 - Exam 2 (Volgushev)

perception

a process in which the brain builds a model of the world, using both previous experience and signals from the sensory organs

awareness of objects and events -> meaning

visual perception

an experience-based interpretation of retinal images

primary visual pathway

retina --> thalamus (Lateral Geniculate Nucleus, LGN) --> cortex (vision, visual perception, object recognition, etc.)

ALSO: retina --> superior colliculus (eye movements)

primary visual cortex, AKA...

V1

striate cortex

Brodmann's area

the image in the eye

reduced and reversed

upside down and right to left

orderly, point to point

retinal image

output: retinal ganglion cells (sampling of image)

dense in center (small receptive fields)

poor in periphery (large receptive fields)

cortical magnification

expanded representation of the foveal (center) stimulus

a small area in central visual field is represented in a larger area of V1 than a large area in the peripheral vision

visual acuity depends on...

retinal location

fovea

highest density of receptors (cones)

lowest convergence (1:1)

highest cortical magnification

highest acuity

periphery

lower density of receptors (rods, cones)

high convergence (many:1)

low cortical magnification

acuity is lower

individual variance

people with higher foveal magnification have higher acuity

functional consequence of more tissue --> higher computational power --> better cortical processing --> better function

retinotopy

points of the visual world that are represented next to one another on the retina are represented next to one another in the visual cortex

orderly representation of the visual world in the cortex

receptive fields of adjacent neurons are systematically shifted in space

oversampling

each point in visual space is covered by thousands of receptive fields

microelectrode recording

Records the activity of a single cell

retinotopy

oversampling

metabolic markers

metabolic radio-isotope marker (C14, 2D glucose)

stimulation

active neurons consume more energy and accumulate more marker

pattern is revealed after processing tissue

drawback: postmortem, one per animal

optical imaging/fMRI

in a living animal, mapping representation of several features possible

measures changes of levels of the oxygenated hemoglobin in brain regions

after onset of grating motion:

- visual cortical neurons fire within milliseconds

- vascular response about 2 seconds later (BOLD negative, cerebral blood flow volume positive)

does not directly measure activity of neurons, but BOLD, due to neurons' activity

fMRI: stimuli

to study representation of the visual space on the cortex: alternate S1/S2, calculate the difference of responses

complementary stimuli in space activate complementary regions of visual cortex

fMRI: results

systematic representation of the visual space on cortex

neighboring points in visual field are represented at neighboring cortical locations

cortical magnification

consequence of retinotopy

stimuli at specific retinal locations activate specific locations in the brain

lines of activity along visual pathway

scotomas

scotoma

regions of absent vision

damage to brain region is reflected as scotoma

- of a certain portion of the visual field

- of a specific shape

possibility to attribute damages along the visual pathway to specific scotoma

receptive field

region on the retina in which visual stimuli influence the neuron's firing rate

lateral geniculate nucleus (LGN)

A structure in the thalamus

stronger surround inhibition

modulation from other brain structures

visual cortex neurons

neurons in the visual cortex extract specific features of the object (orientation) which are then used by the brain in the process of perception

orientation selectivity

emerges in visual cortex

neurons respond to certain orientations but not others

brought about by cooperative action of multiple mechanisms

parallel to perception

End-stopping or hypercomplex receptive fields

receptive fields in the visual cortex as detectors of: edges and strips, oriented bars of specific length

simple cell response

responds to oriented bars at specific location

complex cell response

responds to oriented bars within a certain area

Hubel and Wiesel

studied feature detection in visual cortex and discovered simple, complex, and hypercomplex cells

columns in visual cortex

modular organization

receptive fields of vertically located neurons have similar

- location in visual field

- preferred orientation

- dominant eye

intrinsic signals

distribution of the levels of oxygenated hemoglobin (BOLD) over the cortex --> a correlate of activated regions

orientation map

from single-condition maps, at each location, calculate preferred orientation (strongest response)

ocular dominance map

pool single-condition maps for all orientations for the right and left eye, calculate preferred eye

orientation columns

different orientations are systematically represented in the visual cortex

neurons responding preferentially to same orientations

ocular dominance columns

neurons responding preferentially to the same eye

hypercolumn

complete set of orientations for both eyes

blobs

groups of neurons within V1 that are sensitive to color

interblobs

cells between blobs that are sensitive to orientation and not wavelength (motion and spatial structure)

How does representation of the visual world change from the retina to the cortex?

retina: image

visual cortex:

- representation

- feature maps (retinotopy, oculodominance columns, orientation columns)

selective rearing

experiment in which kittens were reared in an environment of vertical stripes to determine the effect on orientation selectivity of cortical neurons.

- feature detectors are formed in development

- perception depends on neurons

- perception is determined by experience (what we see is what we have learned)

- horizontal reared cats only see horizontal orientation

critical period

representational maps are formed during development

experience is necessary for normal development of the brain (including structure)

experience is necessary for normal perception

plasticity

perception can be trained (within genetic and developmental limits)

magnocellular system

~10%

large cells

large receptive fields

phasic responses

fast axons, 30-40 m/s

movement, contrast, depth

"colorblind"

(Y projection to V1)

dorsal stream, WHERE/HOW

parvocellular system

~80%

smaller cells

small receptive fields

tonic responses

slower axons, 15-23 m/s

texture, color, shape, depth

"motionblind"

(X projection to V1)

ventral stream, WHAT

heterogeneous cells

~10% remaining cells

gamma cells

small somata

large dendritic fields

large receptive fields

tonic/phasic

slow axons, 2-18 m/s

analysis of movement and direction of movement

projections: subcortical structures

function: unconscious reaction, visual reflects

contralateral eye

the eye on the opposite side of the body from the LGN

ipsilateral eye

The eye on the same side of the head as the structure to which the eye sends inputs.

LGN layers

- right/left eye

- parvo (3-6) and magno (1-2) cellular

- retinotopy

layers of visual cortex

1-6

fibers

small pyramids

stellate cells

large pyramids

white matter

inhibitory interneurons throughout

systematic projection of parvo and magno systems from LGN to cortex

L 2/3 --> other cortical areas

L 5 --> callosal, colliculus suerior, pulivinar, pons

l 6 --> LGN (thalamus)

discrimination task

which part of the brain is responsible for the processing?

ablation method

Ungerleider and Mishkin: WHAT and WHERE pathways

WHERE pathway

parietal lobe

landmark discrimination

where/how

dorsal pathway

magnocelllular

WHAT pathway

temporal lobe

object discrimination

what

ventral pathway

parvocellular

one function of V1

sorting information to channels and streams for further selective processing

which part of the brain is responsible for a specific function?

- animal research: ablation method

- observation of patients with localized brain damage (neuropsychology)

- psychophysical experiments: dissociated tasks

double dissociation

mechanisms are different and independent

how to find function with neuropsychology

patients with localized brain damage

tasks: matching orientation vs posting card through slot

- ventral pathway damaged, selective impairment of matching orientation (WHAT) but not of action (WHERE/HOW)

- dorsal pathway damaged, impairment of WHERE HOW

different aspects of visual processing are performed by different cortical areas

psychophysical experiments: dissociated tasks

adjust to match orientation of stimulus with shifted frame vs grasping

visual illusion does not affect grasping

different aspects of visual processing take place in different areas

agnosia

the inability to recognize familiar objects

color agnosia (A18)

color anomia (A18-37)

akinetopsia (mediotemporal bilateral)

object agnosia (A 18,20,21 left and c callosum)

prospagnosia (A 20, 21 bilateral)

representation of faces

neurons process sensory information by extracting more and more complex features and combinations

specific representations of biologically most relevant features

grandmother neuron (or at least highly specialized neuron detector)

distributed representation

weighted activity of a few neuronal groups

high capacity, high reliability

but template coding is faster

representation of objects

extraction of features (translating images into feature codes)

assembling these codes into a percept

- distributed

- partially overlapped

- combination of feature codes

binding problem

which features belong to one object?

possible solutions:

- grouping by timing

- high frequency oscillations

- different synchronized ensembles represent the respective percept

gamma oscillations

faster and more accurate recognition (of objects) correlated with the increased synchrony of EEG across visual and auditory areas

relevant for perception and multimodal integration

features

most common (space, orientation, color)

biologically more relevant (elements of faces)

specificity code

the idea that an object could be represented by the firing of a specialized neuron that responds only to that object

limited capacity, low reliability

what the retina tells the brain

distribution of contrast differences (in lumination and color)

what strategy does the brain use while assembling feature codes?

structuralism vs Gestalt psychology

Gestalt psychology

the whole is different than theh sum of the parts

interpretation is more than a sum of parts (link to top-down processing)

honor physics and avoid accidents

heuristic, mental shortcut

help to make fast decisions about most probably arrangements of objects which produced that image

structuralism

sensations add up to create perception

figure-ground segregation

The perceptual separation of an object from its background.

objects may overlap and their borders may be hidden

knowing the rules and objects help

color helps

neuronal responses as early as V1 can distinguish the figure from the background

viewpoint invariance

objects may be viewed from different perspectives and have different size but are still perceived the same

accidental viewpoint

a viewing position that produces some regularity in the visual image that is not present in the world

image may be ambiguous, allowing more than one interpretation

contrast borders (what retina tells brain) do not necessarily correspond to object borders

defining object borders

results of cortical processing (figure/background segregation) trump results of retinal processing (contrast borders)

how to study an illusion

- quantitative study of an illusion

- clear design of psychophysical experiments

- masking methods

quantitative study of illusions

forced choice paradigm - was shape fat or thin

with masking

variables:

- angle

- distance between elements

- presentation time

measure:

- probability of correct response

- detection threshold

1. illusory contours detected almost as good as real

limit of spatial

2. variable distance between elements - limit of spatial integration

3. variable presentation time

4. backward masking

masking

for perception of illusory contours, ~100 ms presentation is sufficient

interrupts afterimage

backwards masking

perception of illusory contours requires additional ~140-200 ms of processing time

interrupts afterimage and processing

grouping

knowing the occluding object helps

seeing the obstacle helps to recover a hidden pattern

mid-vision

- bring together what belongs together

- split asunder what does not belong together

- use what you know

- avoid accidents and ambiguity of interpretation

perceptual grouping, figure ground segregation, gestalt

Gestalt rules

grouping or organization is not in the stimulus

Law of Proximity

Law of Similarity

Law of Good Continuation

Law of Closure

Also: parallelism, symmetry, meaningfulness, common fate (motion)

formation of perceptual unit characterized by

- superadditivity (whole more than sum of parts)

- transposability (whole preserved despite large changes to parts)

not stimuli alone, but active mechanisms of their processing are involved in perception

figure

in front of the ground

owns the border

is more memorable

- lower area more likely perceived as figure, no difference right/left

- meaningful part more likely perceived as figure

brightness perception

dependent on context and experience

high-level vision

recognition: categorization or matching stimulus representation with representation in memory

recognition and brain activity

recognition

global before local

knowledge of what to expect enhances recognition of the context-appropriate details

global superiority effect

large-scale objects first, then details

the gist of a scene perceived first

contextual influence on recognition

scene, stimulus, mask, question

stimuli which are appropriate to the context are recognized better than those which do not belong to the context

recognition and brain activity

activation of face fusiform area when Harrison Ford shown

conservative conclusion: brain activity correlates with recognition

provocative conclusion: sufficiently high brain activity is necessary for correct recognition

delayed matching to sample experiment

is the test stimulus from same category as sample?

neurons in different brain regions accomplish different tasks

- inferotemporal neurons identification (perception)

- prefrontal neurons preserving information during the delay

is it possible to tell what a person is seeing by measuring brain activity?

voxel experiment - corresponding brain activity with images

72-92% of cases image was correctly identified

sensory stimuli produce unique patterns of brain activity, leading to specific perceptions

prediction: best fit from a pre-defined set

templates

fast

easy to categorize

help to segregate objects

but...

...for viewpoint invariance you need a lot of them

...one neuron one object code is not efficient

image description models

advanced template approach

- templates are created during experience

- more templates for more familiar objects

- viewpoint invariance does not always work, but for most familiar objects it does

experiment: learning letter like figures - no viewpoint invariance

recognition by components theory (RBC)

volumetric features

geons (geometric ions):

- non accidental properties, NAPs

- easily recognizable from most viewpoints (viewpoint invariant)

- simple formalization of description

36 geons - basic shapes

but...

...poor performance with line drawings

...letters

...classes of objects

perception is not always viewpoint invariant

experiments with letter like figures/greebles: poor recognition from new viewpoints (more errors with more deviation from learned viewpoint)

inverted faces, figure ground segregation of familiar contours

support for image description models

multiple committees

categorization:

- entry level category

- subordinate and superordinate categories

- different brain areas (several processes in parallel)

with training, entry level shifts down (previous subordinate becomes entry)

representations in ventral visual object pathway

- small number of category specific regions in addition to a more general purpose region that responds to any kind of objects

- genetically defined regions: face recognition

--- even if damaged early in life, no substitute

perceptual committees

a general scheme of hierarchical processing

feature demons --> cognitive demons --> decision demon

lateral inhibition in neurons and perception

space: antagonistic receptive fields

color: opponent mechanisms

orientation: cross-orientation inhibition

perception: one at a time interpretation of a retinal image

winner take all mechanism