Perception Flashcards
Superior Colliculus
- Parts of circuits for visual processing
- Deploys eye movements
- Mediates target selection
Visual info encoded in V1? and MT/V5?
V1 - Primary visual cortex, receives all visual input
MT/V5 - Detection of motion
Luminace
- the measure of energy emitted/reflected by a light source
- measured in candelas per square metre of surface
Pelli-Robson created
the contrast sensitivity chart
letters on a grid fading
How is contrast coded?
Linear receptive field - Light falls on the retina –> coding of this information leads to excitation or inhibition of neuronal responses
Depth perception
- Humans are very good at discriminating position of an object in depth
- Multiple cues can be used (can be binocular or monocular)
Motion Parallax`
- Relative motion is created when we move
- Close objects move fast, far objects move the least
- We can use this cue to judge the depth of an object
Accommodation
- Accommodation allows one to change focal length
- Lens is stretched or relaxed to bring a target into focus
- Ciliary muscles push or pull the lens
- Strain on the lens is sensed by the visual system and can be used to calculate distance of an object
Accommodation & ageing
With again, the lens becomes less flexible
With less flexibility, it becomes harder to change focal length
Vergence
- Angle of gaze of 2 eyes
- Allows one to change focal distance
- Convergence: yes turned towards each other
- Divergence: eyes turned outwards
Strabismus
Strabismic amblyopia is due to the misalignment of one eye.
Often caused by different muscle tension in each eye
Exotrope - outward pointed eye
Esotrope - inward pointed eye
Often treated with early surgery
How is stereo detected in the brain?
- Cells in the early visual cortex are binocular - they respond only when something is in the field both in LE view and the RE view
- Cells in the MT code disparity
What is stereo good for?
Sewing
Picking berries
Surgey
Goal-directed movements
- made to interact with our environment
- Goal often visually defined
- outcome often important
- e.g. swatting a fly/surgey/picking berries
Goal-directed movements: Rapid and Slow guided movements
Rapid - initial visual information significant in determining pointing performance, little time for online correction
slow, guided movements - A lot of time for online correction
To make a goal-directed hand movement
- localise a target in space
- Formulate a plan to move the hand
- execute the hand movement
Control of hand movement: 2 phases
- The planning phase: Initial visual estimate of location, plan a course
- The guidance phase: Observers are comparing their visible hand trajectory to an invisible planned trajectory
Fitts’ Law
- the faster the movement, the less accurate the response will be
- Speed/accuracy trade off
Fitts’ Tasks
- Repetitive, simple tasks
- Varied amplitude and information content of tasks
- measured time to compete and error
- Results: found that performance varies as function of amplitude and tolerance requirements
Motion. Helps:
- Figure-ground segmentation
- Extraction of 3D structure
- Visual guidance of action
Barlow & Levick (1965)
- Rabbit cells in retina that were directionally selective
- Suggested a building block for motion models
- Results indicated that the activity of single neurons could be related to functional behaviour
Simple motion detection unit
Taking 2 samples from 2 locations in space, with a time delay in between.
Unit (X) receives input from 2 spatially separated cells, with a temporal delay between them.
models of this type first proposed by Reichardt (1961) based on insect work
Reichardt detector
- simple space/time plot
- getting one discrete location and comparing them with a time delay
- A specific form of neural motion detector, which combines signals initiated at slightly different times from adjacent retinal locations
- Contains a bank of temproal and spatial filters
- comparison of these outputs used to detect motion
Energy model
- space/time plot
- suggests that a critical stage of motion processing is the response of linear filters that are oriented in space and time and tuned for spatial frequency
- Similar banks of filters as Reichardt model, but now has outputs from quadrature pairs that are summed to calculate motion energy.
what does this extra stage buy you? Oriented linear responses
Benefits of energy model
- no need for discrete correspondence
- specificity - motion response is to a particular space, time & spatial frequency
- More flexible definition of a feature
- has the ability to detect motion in random noise patterns
Aperture problem
- motion of a repetitive pattern is perceived as moving in an ambiguous direction
- a grating behind an aperture appears to be moving in the same direction as the orientation of the longest side of the aperture
Local vs. global features
- local receptive fields do not have enough info to disambiguate direction of the object motion
- Local direction of any line in an aperture is consistent with many different line directions
An ON centre - OFF surround receptive field organisation can account for
Lateral inhibition: capacity of an excited neuron to reduce the activity of its neighbours
Westheimer and McKee (1977) reported on human localisation thresholds for two lines. These findings were remarkable because they demonstrated that localisation thresholds were
Less than the width of a photoreceptor
Solving the aperture problem
- Orientation information can be used to disambiguate edges
- Changing the orientation of the aperture changed perceived direction of motion
- pooling orientation and motion information over many spatial scales aids disambiguation of motion detection
Motion deblurring
- Cells in V1 integrate over ~100ms
- Results in blur among the axis of motion for moving objects
- However, humans do not see blur, which suggests that motion delblurring mechanisms exist.
- Effectively creates an orientation signal along the axis of motion
- This form information can then be used to disambiguate direction of motion
Biological motion
Johansson (1963) demonstrated that people are able to recognise people walking, even when there are only points of light on the joints
Visual and motor error
- Hand precision and visual precision must be nearly identical in tasks where movements are slowly and continuously guided –> e.g. surgery, paining, embroidery
- precision is limited by random noise
Internalised reference frameworks
E.g. Phone example - point to the camera icon, how do you know which one it is?
- we use internalised reference frameworks. Based on a quick glance at the surroundings and memory, it allows us to make rapid movements in familiar settings
How to minimise visual error
Dual model of movement control
??
Solving the aperture problem
- Orientation information can be used to disambiguate edges
- Changing the orientation of the aperture changed perceived direction of motion
- pooling orientation and motion information over many spatial scales aids disambiguation of motion detection
Motion deblurring
- Cells in V1 integrate over ~100ms
- Results in blur among the axis of motion for moving objects
- However, humans do not see blur, which suggests that motion delblurring mechanisms exist.
- Effectively creates an orientation signal along the axis of motion
- This form information can then be used to disambiguate direction of motion
Biological motion
Johansson (1963) demonstrated that people are able to recognise people walking, even when there are only points of light on the joints
Visual and motor error
- Hand precision and visual precision must be nearly identical in tasks where movements are slowly and continuously guided –> e.g. surgery, paining, embroidery
- precision is limited by random noise
Internalised reference frameworks
E.g. Phone example - point to the camera icon, how do you know which one it is?
- we use internalised reference frameworks. Based on a quick glance at the surroundings and memory, it allows us to make rapid movements in familiar settings
How to minimise visual error
Dual model of movement control
??
Acting in a changing world
People make rapid hand movements to interact with their environment, however, things don’t always stay in the same place for long
Online control of movement
Online control is the alteration of the motor plan on the basis of a discrepancy between predicted and real feedback.
Are movements ballistic?
No, we can use visual information throught the reach to update performance
Common eye movements
Saccades
Smooth pursuit eye movements
Saccades
- 3 times/second
- Movement characterised by acceleration to a new location.
- allow us to foveate points of interest
Smooth pursuit eye movements
- Tracking movements
- Movements characterised by constant velocity once at same speed as target
- Allow us to track item of interest (often seen with a saccade)
Two visual streams of processing
DORSAL ACTION STREAM -
retina -> LGNd -> V1 —DAS> Posterior Parietal Cortex
VENTRAL PERCEPTION STREAM
retina -> LGNd -> V1 —VPS> Inferotemporal cortex
How do people know where objects are?
- Visual localisation
- From experience with the world
- Information feedback from eye movements and hand movements
- People have a dynamic representation of space that is updated as they move around
Developmental Coordination Disorder (DCD)
- ‘clumsy’ children - have trouble with fine motor control (sport, writing)
- Mechanisms underlying DCD not yet entirely clear
Parietal damage
Often caused by stroke
In some cases, leads to visual neglect (difficulty with daily living)
Reasons to move our eyes
- Limited spatial resolution in the periphery
- Shifting the eyes shifts the fovea: a way to gather high (er) resolution about the environment
- Gather information for perception and for action
Visibility maps
Maps of visual acuity across the field
Used to guide eye movements & probably hand movements
Eye-hand coordination
- Eye & hand movement usually spatially correlated, and usually temporally correlated
- During rapid pointing - eye typically leads the hand
Yarbus (1961)
Found: eye movements (saccades) are NOT random
Eye movement strategies differed depending on the question
Conceptualising attention- key models
- Signal detection theory based accounts
- Feature integration based accounts
- Salience based models
How do we select a new target?
Super colliculus
The concept of salience
- Target that is ‘different’ to its surroundings
- attracts attention
Hierarchy of processing
- receptors
- neurons transmit information
- thalamus (relay station(
- cortical processing
A ‘receptive field’ is
the area over which a cell changes its activity in response to a change in stimulus (e.g. luminance or colour)
Electrophysiology
- Recording electrical activity from a single cell using fine insulated wires
- allows to measure function of cell in real time
- Strength of single cell recording is its excellent spatial and temporal resolution
- Weakness is that one is recording from a very small area of the brain
Dorsal & Ventral pathways
Dorsal - ‘where’
- LGN –> Parital lobe
- Motion processing
- Spatial processing
Ventral - ‘what’
- LGN to temporal lobe
- object processing
- Fine detail
Maps in the brain
Representation of stimulation is organised in the cortex to mirror stimuli in the world
- Retinotopic maps - In the cortex, preserve the relative position of stimuli as they are received on the retina
fMRI
- measures change in blood flow in real time as participants complete a task (e.g. watching a computer screen and pressing a button to respond to an attentional task)
- strength - include that the entire brain can be imaged, and therefore that its a real time measure of activity in different areas in the brain
- weakness - limited temporal and spatial resolution
Psychophysics
- Experimental techniques for measuring the percept associated with a stimulus of a given intensity
- looks at physical stimuli and the sensations and perceptions they affect
psychophysical techniques:
- method of constant stimuli
- staircase procedure
- two alternative forced choice
- method of limits
- method of adjustment
Hyperacuity
- Measured sensitivity to position using psychophysics
- Found that the ability to discriminate was better than width of a photoreceptor
Disparity information
- Coded in MT
- Different vision in each eye - different views of the world - relative position in space
- further object less different in eye
Magnocellular pathway
- motion
- large cells compared to parvocellular
Parvocellular pathway
greater spatial resolution, but lower temporal resolution than magnocellular
- smaller cells
Gestalt movement
Apparent motion
- brain carried out some kind of inferential processing on information from the environment
