PS222 Perception F25 Exam 2

retinal ganglion cells RGC

• organize into into light-dark boundaries

• deemphasize uniformity

• emphasize boundaries

• behaviorally relevant units

receptive field size

determined by how many photoreceptors (?) converge at at retinal ganglion cell

• dendritic trees show single RGC gets signal from many bipolar cells

• NOT UNIFORM across retina

parasol RGCs

more convergence

→ lots of branches

→ signals from rods,  PERIPHERY

→ projects to M layers in LGN

small RGCs 

less convergence 

→ info from cones

→ M/L waves 

→ projects to P layers in LGN

small bistratisfied RGCs

moderate convergence

→ info from cones

→ S waves

→ projects to K layers in LGN

fovea

• high acuity/sharpness

• small RFs

• see more details

periphery

• high sensitivity

• larger RFs

• likely to detect faint stimuli

retinotopic mapping

next to each other in the retina/LGN (upside down, left-right reversed) means next to each other in the real world

right visual field

nasal right, temporal left

left visual field 

temp right, nasal left 

ipsilateral fibers

temporal side, stays on the same side

contralateral fibers

nasal side, crossover

optic tracts 

past crossover (optic chiasm), rep the same part of the visual field 

left side of the visual field →

right side of the brain

• axons from temporal leave eye and stay on same side of brain

aka TEMPORAL = SAME SIDE

superior colliculus (10% axons)

• deep in brain

• phylogenetically older so found in species we not closely related to

• where > what

• important for eye movement planning i.e tracking

• multisensory with RFs in many sensory areas (i.e moving and making sound)

lateral geniculate nucleus LGN (90% axons)

• RGC axons terminate in diff layers of lateral geniculate nucleus LGN 

• integrate info from 2 eyes, further organize

magnocellular layers

layers 1, 2

• parasol RGC

• periphery

• rods

parvocellular layers

layers 3, 4, 5, 6

• small RGC

• fovea

• cones

koniocelllular layers

between layers 3, 4, 5, 6

the pink between the brown layers

• small bistratisfied RGC

• fovea 

• cones

M-cells

strong opponency - LGN emphasizes light-dark boundaries

• detects fast change, motion, flicker

• bigger RFs

P-cells & K-cells

• color component cells

• smaller RFs

V1 cells’ receptive fields organized

retinotopic and columnar (orientation and hypercolumns)

• hypercolumns consist of orientation columns and ocular dominance columns

change in orientation of stimuli

changes activity from baseline

tuning curve

the peak signifies the preferred orientation of the neuron’s RF

what does v1 cells need to become orientation-selective?

needs to be constructed from other neurons

simple cells 

well-defined organization of on/off regions 

cats vs primates

(hubert) cats: mainly simple cells in V1

(later studies) primates: mainly complex cells in V1

complex cells

respond to a stimulus anywhere in the RF

• no clearly defined on-off regions

responds to orientation

both simple and complex cells

binocular cells

respond to input from both eyes

color-sensitive cells

respond to edges of a particular color

input from LGN

• light dark boundaries across visual field

• integrated information from both eyes

• retinotopically organized

cells in V1 organized so that…

• any orientation can be detected anywhere in the visual field 

• info from both eyes can be integrated 

• info is still retinotopically organzied 

orientation columns (V1 organization)

neurons tuned to the same orientation

ocular dominance columns (V1 organization)

get signals from both eyes (one or the other and then combines)

• binocular vision 

hypercolumns

orientation & ocular dominance columns

• occurs all over V1

• allows us to detect orientation at any depth

• columns: diff orientations

• takes input from both eyes

cortical magnification (V1 organization) 

objects not always represented accurately/the same 

• fixation point in visual field turns out to be larger in the retinotopic map in V1

incoming visual information organization LGN → V1

LGN:

• light dark boundaries

• color, motion, direction, speed

• information from the two eyes 

V1

• edges of particular orientation, length and width

• color 

• motion direction and speed

• depth 

dorsal pathway

“where/how”

• MT: motion

• parietal cortex: perceiving space and motion + coordination visual-motor interactions 

ventral pathway

“what”

• V4: form & color (curvature)

• inferotemporal cortex: object recognition

• LOC lateral ocipital cortex

double dissociation

legion impairs skill A but leaves skill B intact; each skill has diff functions

legion in parietal cortex

failed landmark task (“where“), succeeded object task (“what”)

legion in inferotemporal cortex

failed object task (“what”), succeeded landmark task (“where”)

perceptual matching

doesnt need to put into anything, just match the orientation by rotating it

posting

put the envelope in the mailbox; identify where the slot is

LOC lateral occipitial cortex (ventral)

responds to whole form

• larger RF

IT inferotemporal cortex (ventral)

specific objects

• cells highly selective 

• sensitive to small changes affecting object categorization

object agnosia (IT, ventral)

lesions cause impaired recognition

fusiform face area FFA

inside the inferotemporal cortex

• brain activates when you look at faces

• responds to visual expertise too — expert perception area

prosopagnosia / face blindness

lesions to FFA result in impairment of face recognition

face processing

upright faces processed holistically

inverted faces processed by parts 

MT middle temporal cortex (dorsal)

motion, prefer moving stimuli, particularly direction and speed

IPS intraparietal sulcus (dorsal)

visually-guided motion

1. LIP: eye movement

2. MIP: reaching & grasping

3. PRR: parietal reach region (primates only)

4. AIP: active during manipulation of objects (i..e fingers)

problems faced by visual system

1. image clutter: necessary byproduct of living in a 3D world

2. object familiarity: need to recall what we dont know

3. object variety

4. variable views: not always the same vantage point

object recognition steps

represent objects

1. detect different levels of detail

2. perceptual organization

recognize objects

1. match perception to long term memory (cognition)

perceptual organization (object recog)

• find form aka edges and curvature

• foreground vs background

• fill in missing parts (what you cant see)

object recognition 

pattern of neural activity driven by physical stimuli in the world 

spatial scale

range of levels of details in an image (course → fine)

course details

LOW spatial frequency

• # of changes in contrast is low

fine details

HIGH spatial frequency

• # of changes in contrast is high

just high spatial frequency

sharp, difficult to recognize object and hard to segment

just low spatial frequency

blurry

contrast

helps see different spatial frequencies

• middle frequency — most sensitive

contrast sensitivity function

• above the curve → gray

• at the curve/below aka absolute threshold→ still just barely detect light + dark

• varies across species 

perceptual interpolation

make conclusions based on what we see

neural basis of edge and curvature representation (percep. org.)

V1 cells (orientation) and V4 cells (curvature) 

• sent to LOC and IT

neural basis of figure-ground assignment (percep. org.)

V2 receptive field

• right darker bg, left lighter bg

• light oriented edge

• responds more when figure is on the left

neural basis of perceptual interpolation (percep. org.)

still works for partially occluded objects 

top-down influences of recognition

knowledge & expectations matter

neuronal peer pressure

listen to neuron buddies getting signals from similar or adjacent areas on the retina

newton’s insight

pure lights → can’t be broken into composite colors

white lights → mixture of pure lights (7)

color (on color solid)

hue, saturation, brightness

• no discrete boundaries between colors 

hue

distinguishes different color categories

→ AROUND the wheel

saturation

dominance of hue in the color

→ INSIDE-OUTSIDE of the surface of the wheel

brightness

perceived light emitted

complementary colors

colors that

• are on opposite of each other on the color solid

• when added together, appear white or gray (natural light?)

blue green red BGR (beggar)

• short medium long (wavelengths)

• 400 - 700 

2 physical properties of perceived color

1. reflectance properties of surface (absorb vs reflect)

2. spectral quality of illuminant 

reflectance properties of objects

white paper — reflects the most

black paper — absorbs all uniformly

spectral quality of illuminant

how much light/what wavelengths are refleced ALSO depends on the light source

how to detect physical stimuli?

photoreceptors

• rods, cones (small medium and large)

• blue-green and orange-red have same receptor response

univariance principle

need more than 1 cone to distinguish colors, diff wavelength intensity combinations can elicit the same response

• NEED AT LEAST 2 CONE TYPES TO DISCRIMINATE COLOR

trichromats

(i.e. humans) the types of cones you have dictate the colors you can detect and discriminate 

• blue peak: 430 nm

• green peak: 530 nm

• red peak: 560 nm

all diff color together 450 nm light?

cone types variation

the more cone types, the better your color discrimination 

• colorblind: 1 cone type

• mammals less cone types, insects more cone types

2 representational stages of color perception

1. trichromatic color representation

2. color opponent representation 

color opponency

visual sys treats certain colors as opponent pairs → we see afterimages; depends on how RGC hooks up to cones

• red-green

• blue-yellow

cone fatigue

tired from continuous response

cone bleaching

stops responding to light it likes to respond to

color constancy

an object’s perceived color remains constant despite changes in light falling on that object

compensation

visual sys turns down neural responses to wavelengths that are disproportionately abundant