The seeing brain

The vision problem

• vision is computationally complex

• Sensory input is ambiguous (more than one meaning)

• Different perceptions can arise from the same input

• brain processing, not just sensory input

The inverse problem

→ challenge of converting 3D image of the world into a 2D retinal image

• Ambiguity = many different configurations of a surface could lead to a single retinal input

-requires context, prior knowledge and assumptions

• Gestalt e.g. Necker cube = sudden involuntary perceptual change where the brain flips from one interpretation of an ambiguous image to another without the image itself changing

Sensation vs perception

Sensation = the raw input

• when a stimulus (light, sound etc.) activates our sensory organs

Perception = the brains interpretation

• using past knowledge to organise and make sense of that sensory input

The retina

→the light sensitive inner surface of the eye, made up of multiple layers of:

Photoreceptors = convert light to neutral signals

Signal flow within the retina = photoreceptors → intermediary neurons → ganglion cells

Types pf photoreceptors

• rod cells - specialised for low light intensity

• Cone cells - specialised for high light intensity

• Fovea - highest conc of cones and visual acuity

Receptive field

Region of visual space a neuron responds to

Early feature detection

Early visual neurons respond to:

• light vs dark regions

• Edges and orientation

• Motion direction

→ early visual areas extract simple features that form building. Locks for later more complex visual processing

Lateral inhibition - early detection

• neighbouring neurons inhibit each other

• Enhances contrast

• Sharpens edge detection

From the eye to the brain

• Info leaves each eye via Optic nerve

• then branches to multiple brain targets

• Main and best studied pathway (about 90% of fibres) = optic nerve → lateral geniculate nucleus (LGN) → primary visual cortex

What is V1

• primary visual cortex

• Retinotopic organisation

• Basic visual features processed here

Early visual processing

• V1 sends info feed forward → higher order visual areas

• These include V2, V3, V4 and motion sensitive area MT

Examples of specialised visual regions:

• fusiform face area (FFA) = face perception

• Lateral occipital cortex (LOC) = object recognition

• Parahippocampal place are (PPA) = scene and place perception

Dorsal vs ventral stream

Dorsal = ’where/how’, spatial awareness and motion

Ventral = ‘what’, object recognition

The role of feedback between high and low

• for visual awareness to emerge, it needs feedforward and feedback between higher order and lower order cortical areas

• Feedforward is automatic and leads to abstract and categorical representations in higher cortical eras

• Feedback is needed to analyse visual information in a more detailed manner

What does object recognition depend on

• combining simple features into complex shapes

• Matching input to stored memories

• Feedback signals that refine and disambiguate perception

• Experience shaping and tuning visual representations

→Object recognition is an active, experience-dependant process

Where is object recognition

Relies on an interconnected network:

• Inferior temporal cortex (IT) = recognising complex objects/shapes

• Perirhinal cortex (PR) = links visual features to memory and object identity

• Parahippocampal cortex (PH) = recognising scenes/places

• Prefrontal cortex (PF) = decision making, categorisation, task goals

Factors that influence object recognition

• environment = perceptual constancy - the visual system has to identify the same object under different environmental conditions

• Intrinsic variability = categorical perception - objects within same category show some level of variability

• Previous experience = associative recall - creation of object templates to compare new objects to

What is perceptual constancy

→ ability of our brains to see objects as stable and unchanging despite dramatic shifts in sensory info hitting our eyes

When is an object perceived to be the same

1. SIZE - The size of its image on the retina becomes larger or smaller due to the viewing distance

2. POSITION - The location in the retinal image changes

3. FORM-CUE - There are changes in reflectance e.g. changes in properties that define it (colour/texture)

What is a viewpoint

• structural descriptions: 3D memory representations of objects

• Principal axis: usual viewpoint representation of an object

• Right parietal lobe: sensitive to viewpoint

• Left IT: insensitive to viewpoint

• Two parallel routes for perceptual constancy

Categorical perception

• ability to perceive different objects as the same category, even if they differ e.g. apples in a fruit bowl

• Category-specific responses are common in neurons of the lateral prefrontal context, which receives direct info from IT

Perceptual learning (experience)

• object recognition is modified by experience

Two forms of learning:

• Implicit = improve at recognising objects without conscious effort

• Explicit = consciously remember and identify new objects

Mediated by response properties of inferior temporal (IT) neurons

Associative recall

→ cognitive process of retrieving a memory triggered by a related cue or stimulus

• contribution of inferior temporal cortex (IT) to object recognition can be modified by pervious experience

• Consolidated by inputs from memory structures in the temporal lobe

• Activation of working memory from the prefrontal cortex can also mediate this effect

Face recognition

• relies on different neural substrates to the recognition of other objects

• System has to recognise the identity of a face, in addition to its changeable aspects (expresison)

• Mediated by a distributed bilateral neural system located in the occipitotemproal extrastritae visual cortex

Greebles

face selective neurons also respond to other categories of objects e.g. artificially generated objects called Greebles

Visual awareness

• regions involved in object perception become more active when we are aware of that object

• Not one single brain area responsible

• Activity evoked by a stimulus in a specific brain area is necessary, but NOT sufficient to generate a conscious perception of that stimulus

• visual awareness results from distributed interaction between different lower-order (sensory) and higher-level associative regions

Different disorders of vision

Damage to different visual regions produces characteristic patterns of deficits: V1 → cortical blindness, blindsight

Dorsal stream → hemispatial neglect

​

Damage to V1

• Complete damage to one side of V1 = cortical blindness for one side of space (hemianopia)

Partial damage can lead to specific areas of blindness:

• upper part of V1 = bottom part of space

• Lower part of V1 = top part of space

→ quadantanopia

Small region of damage = scotoma

Disorders of V1

Anton’s syndrome

• cortical blindness with anosognosia - complete unawareness of total visual loss in both hemifields

• Lack of concern, denial and confabulation - ansodiaphoria

• Bilateral lesions to V1 likely involving interruption of V1 from association cortices but no definitive cause

Blindsight

• visual perception is preserved, but visual awareness is not

Dorsal stream - hemispatial neglect

• visual scenes contain multiple elements that compete for limited attentional resources

• Spatial attention is modulated by a dorsal attention network, influencing processing even in early visual areas

• Key regions = posterior parietal cortex (PPC), front eye fields(FEF) and cingulate cortex

• Damage to this network, particularly the right hemisphere, impairs attention to the contralateral (left) side of space

• Neglect is not a visual deficit and can affect multiple sensory modalities

Ventral stream - prospagnosia

• inability to recognise familiar faces or learn new ones

• Other senses intact

• May identify individuals based on voice, posture, smell etc

• Able to match faces, distinguish between faces and determine sex, age etc.

• Damage to the lateral occipital cortex and IT area