Vision Notes
The Eye
Key components of the eye:
Ligament
Iris
Pupil
Lens
Cornea
Eye muscle
Ciliary muscle
Sclera (the white part of the eye)
Retina
Fovea
Optic nerve
Blind spot
Receptive Fields
A receptive field is the specific location of a visual image where a stimulus can influence the activity of a given cell.
Photoreceptor receptive fields are determined by their position in the retina and the eye's optics.
Neural Integration: Spatial Summation
Spatial summation is the process by which EPSPs (Excitatory Postsynaptic Potentials) are added up and IPSPs (Inhibitory Postsynaptic Potentials) are subtracted on a neuron.
This integration occurs at different sites on the same neuron, specifically at the axon hillock.
Center-Surround Receptive Fields
Downstream of photoreceptors, in the retina and brain, receptive fields result from the addition and subtraction of photoreceptor receptive fields.
Types of Retinal Ganglion Cells
On-center retinal ganglion cell:
Light in the center increases the firing rate.
Light in the surround decreases the firing rate.
Off-center retinal ganglion cell:
Light in the center decreases the firing rate.
Light in the surround increases the firing rate.
Visual Field Mapping
The receptive fields of many neurons together form a map of the visual field.
Both eyes see both sides of the visual field.
Both retinas project to both sides of the brain.
Axons segregate to ensure that visual brain areas map the contralateral visual field.
Maps as Representations
Maps are representations that preserve neighbor relationships.
The quote "The map is not the territory" emphasizes that a map cannot be a perfect copy; it must distill important elements while omitting others to be useful (Jorge Luis Borges).
A map takes in whats important and leaves out was is not
Topographic Mapping
Topographic mapping is a general principle across sensory systems.
The somatosensory and motor cortex are organized as maps of the body surface.
Auditory cortex is organized as a map of sound frequency.
The apex of the cochlea corresponds to lower frequencies (e.g., ).
The base of the cochlea corresponds to higher frequencies (e.g., ).
Retinotopic Mapping and Cortical Magnification
Visual cortex is organized as a map of the visual field through retinotopic mapping.
Receptive fields of neurons are topographic, meaning nearby neurons respond to nearby parts of the visual field.
Cortical magnification refers to the disproportionate representation of the fovea in the cortex.
The binocular zone is the part of the visual field seen by both eyes.
Visual Cortex Lesions and Scotoma
There's a topographic relationship between visual cortex lesions and the location of scotomas (blind spots).
Calcium Imaging
Calcium imaging uses sensors that change fluorescence intensity when enters the cell.
The sensor is based on a gene from jellyfish called GFP (Green Fluorescent Protein).
Topographic Mapping of Visual Cortex Activity
Visual cortex activity and stimulus location have a topographic relationship.
The visual cortex is organized as a map of the visual field.
Neural Integration
Neural integration involves the spatial summation of EPSPs and IPSPs that occur at different sites on the same neuron.
Two simultaneous EPSPs sum to produce a greater EPSP.
Two simultaneous IPSPs sum to produce a greater IPSP.
A simultaneous IPSP and EPSP can cancel each other out.
Center-Surround Receptive Fields Formation
Center-surround receptive fields are formed by adding and subtracting photoreceptor receptive fields through horizontal cells.
This process involves bipolar cells.
Retinal Ganglion Cells
On-center retinal ganglion cells:
Light in the center increases the firing rate.
Light in the surround decreases the firing rate.
Off-center retinal ganglion cells:
Light in the center decreases the firing rate.
Light in the surround increases the firing rate.
Simple Cell Receptive Fields
Adding up circular receptive fields can create a simple cell with an oriented line receptive field.
Simple Cells in Primary Visual Cortex
Simple cells in the primary visual cortex respond best to bars with a preferred orientation.
The diagram shows how lateral geniculate cells contribute to the receptive field of a simple cell.
'On' and 'Off' areas determine the cell's response to light.
Complex Cells in Primary Visual Cortex
Complex cells respond best to moving bars with a preferred orientation.
Visual Cortex Devotion
A significant portion of the cortex is dedicated to vision.
Visual Cortex Specialization
In blind individuals, the visual cortex can be repurposed for other functions like spoken language, math, and touch.
Modularity in Visual Cortex
The visual system is divided into two pathways: the "what" and "where" pathways.
Different regions of extrastriate cortex process different aspects of the visual image (Van Essen et al., 1992).
"What" and "Where" Pathways
"What" pathway (ventral stream):
Located ventrally, spanning the occipital and temporal lobes.
Neurons are sensitive to object properties like shape and color.
Damage impairs conscious perception of objects.
"Where" pathway (dorsal stream):
Located dorsally, spanning the occipital and parietal lobes.
Neurons are sensitive to spatial properties/motion.
Damage impairs visually-guided behaviors (e.g., reaching, grasping).
Visual Pathways
Ventral (temporal) pathway:
Includes areas like AIT, CIT, PIT, and V4.
Dorsal (parietal) pathway:
Includes areas like 7a, LIP, MST, MT, and VIP.
Parvo and Magno pathways:
P cells form the P pathway.
M cells form the M pathway.
Color and Motion Selectivity
Area MT neurons mostly respond to moving stimuli.
Area V4 neurons mostly respond to the color of stimuli.
Visual Cortex Specialization and Deficits
Damage to individual areas causes specific deficits:
Achromatopsia (inability to see color) – area V4.
Akinetopsia (inability to see motion) – area MT.
Random Dot Motion Task
Monkeys discriminate dot stimuli based on the direction of motion.
Discrimination is easy when the % correlation is high and gets worse as the dots become more random (Britten et al., 1992).
MT Activity and Motion Coherence
Activity in MT correlates with the coherence of the motion stimulus (Britten et al., 1992).
Neuron Perception
Individual neurons can be as perceptive as the whole monkey (Britten et al., 1992).
MT Activity and Behavioral Choice
Activity in MT correlates with behavioral choice (Britten et al., 1996).
MT Stimulation
Stimulation of MT influences the perception of motion (Salzmann et al., 1990).
Adding Inputs and Vision
Adding and subtracting photoreceptor inputs produce a center-surround receptive field.
Adding and subtracting center-surround receptive field inputs produces a simple cell receptive field (oriented bar).
Limitations of Simple Addition Models
It suggests a need for a large number of cells, potentially one for each face.
It raises concerns about the loss of specific representations if a single cell dies.
It necessitates invariant representation of faces.
Jennifer Aniston Cell
The concept of a Jennifer Aniston cell raises questions about how the brain recognizes specific individuals and objects.
Invariant Object Recognition
Invariant object recognition is the ability to recognize objects regardless of changes in viewpoint, lighting, or other variations.
Specialization of Ventral Visual Areas
Ventral visual areas are specialized.
Responses to faces and houses show distinct patterns (Kanwisher et al., 97-99; Tong et al.; Sergent et al., 92; Haxby et al., 91, 94, 99; Puce et al., 95, 96; McCarthy et al., 97; Halgren et al., 99).
Specialized Visual Areas
Fusiform Face Area (FFA): Specialized for face recognition.
Parahippocampal Place Area (PPA): Specialized for recognizing places.
LOC: Processes information about things.
Body Area: Processes information about bodies (Downing et al., 01).
Stanford Program
Stanford Human Intracranial Cognitive Electrophysiology Program studies cognitive functions using intracranial recordings.
Biological Motion
Biological motion involves the simultaneous processing of what and where.
Posterior STS processes biological motion vs. scrambled motion.
EBA processes headless bodies vs. objects (stationary).
LOCd processes faces & objects vs. texture.
FFA & OFA process faces vs. objects.