Adaptation III Flashcards
visual motion processing
What is motion good for?
- Motion based image segmentation
- Navigation
- Depth from motion
- Structure from motion
How does the brain compute visual motion?
Through two methods: indirect method and direct method
IM: independent analysis of spatial displacements and temporal intervals, used when few objects/features, long intervals & large displacements (e.g. Braddick 1974)
DM: specialised detectors compute motion from intensity variations in retinal image without feature tracking. motion perception possible with subthreshold spatial and temporal displacements (Exner 1875), and brief, high intensity displays precluding cognitive feature tracking (Braddick 1980).
Outline principles of motion-sensitive receptive fields
Some V1 cells hace receptive fields oriented in space-time, so strong response to oriented edges moving in preferred direction, but not opposite direction.
Luminance profiles of moving contours are oriented in space-time, so motion detection analogous to extraction of spatial orientation.
What is the Reichhardt model of motion detection?
Receptive fields sample two adjacent points in space, so moving stimulus activates each in turn. Output of first neuron delayed relative to second. Both outputs compared.
Movement left = no response, movement right = motion
What are some limitations of local motion analysis?
Problem with aperture: local motion detectors in V1 see small part of image and respond to motion orthogonal to luminance edges, so output of any motion detector might not be valid indicator of overall moving direction of object.
Speed selectivity: local motion detectors in V1 respond to one combination of SF and TF, rather than all stimuli moving at set speed.
motion integration in MT (middle temporal visual area) compared to V1
many more neurons in MT aso show proper speed tuning than in V1. V1 neurons respond to a particular orientation in a set direction, whereas MT neurons respond to preferred direction independent of pattern. (Priebe, Lisberger and Movashon 2006)
Outline the principles of higher-order motion processing
neurons in MST area (medial superior temporal) have large receptive fields to respond selectively to optic flow rotation, contraction and expansion.
neurons in STS respond selectively to biological motion.
Regions in IPS and LOS implicated in extraction of structure from motion in fMRI studies.
Outline neural consequences of motion adaptation in V1
responsivity of direction selective V1 neurons reduced by repeated stimulation in preferred direction.
adaptation reduces response of some neurons to stimuli at all contrast levels but shifts contrast response function laterally for others. (Crowder et al., 2006)
in MT, adaptation to drift gratings produces a mix of resposes and contrast gain changes in neurons. this disappears when adapting and test stimuli presented in different regions of receptive field - suggesting effects of adaptation are likely inherited from V1.
M1 neurons show initial transient peak in firing rate that decays to sustained level. This reduction in responsiveness is kept when stimuli are moved to different regions of receptive fields, suggesting it isnt inherited from V1 inputs.
what are the perceptual consequences of motion adaptation
changes in visual sensitivity and bias mirroring those following adaptation to static stimuli are observed after motion adaptation.
contrast sensitivity: reduced for gratings moving in adapted direction but not in opposite direction (Sekuler and Ganz 1963)
direction aftereffect: adaptation to given direction of motion produces repulsive biases in perceived direction of following stimuli (Levinson and Sekuler 1976)
The motion aftereffect: the waterfall illusion
after adaptation to a given direction of motion, static or flickering objects appear to move in the opposite direction.
Robert Adams (1834)
How can the motion aftereffect be explained?
Distribution shift model (Mather 1980) - prior to adaptation, static test pattern produces similar responses from motion selective neurons tuned to all directions. Selectively reduces neural responsivity to adapted direction causing a distribution shift in responses to test pattern.
What are the properties of the static motion aftereffect?
Measure duration of illusory motion percept using static test patterns: so motion aftereffect shows some properties consistent with adaptation at early stage of local motion processing in V1.
- Location specificity
- Partial interocular transfer
- Spatial frequency tuning
- Temporal frequency tuning
What is the dynamic (flicker) motion aftereffect?
test motion aftereffect with moving test stimulus and try to null percept of illusory motion.
in absence of adaptation, equal component contrasts gives rise to perception of counter phase flicker (no net motion - participants equally likely to say up or down).
strength of MAE can be quantified by measuring the shift in this null point post adaptation in a given direction.
What are the properties of the dynamic motion aftereffect?
Dynamic MAE shows several characteristics consistent with adaptation at higher stage processing, where local motion signals are integrated.
- Low positional specificity: adapting at one retinal location reduces MAEs at remote spatial locations, particularly when complex optic flow motion used.
- Complete interocular transfer
- Speed tuning
Are there multiple stages of motion adaptation?
can differing properties of static and dynamic MAEs reflect adaptation at local (v1) and global (MT, MST) stages of motion processing respectively?
some questions to answer: different types of test patterns reveal different stages motion adaptation.
do these different MAEs reflect different forms of adaptation seen physiologically.
