As which process is the sensory processing seen?
as inference process
Why does RGCs (retinal ganglion cells) respond strongly when there are intensity differences in their RF & only wekly to unfirom stimulation of the RF?
due to the center-surround RF structure
How ist redundany removed in the visual system?
by removing spatial correlations by lateral inhibition
What is the task of parosal ganglion cells?
show phasic response behavior
repsond transiently to stimulus changes & thus reduce redundancy by removing temporal correlations
What is the task of small bistratified ganglion cells?
specialized for encoding of spectral information
show opponency ans thud reduce redunda by removing correlations in the wavelength domain
When does orientation selectivity happen?
coding property that arises in cortex
What is the goal of sensory coding?
coding with:
minimal activity
minimal redundancy
Which structure does the functions in finding sparse image coding capture?
localized & oriented structure
similiar to receptive fields of orientation-selective neurons in primary visaul cortex
How is redundany reduction done in VS?
Decorrelation
What does zero-phase whitening do?
decorrelates pixelwise
preserving sructure in outputs
What does PCS decorrelation do?
decorallates by rotating to axes of highest variance
rotation matrix E = matrix of Eigenvectors
D = diagonal matrix of eigenvalues
What does ICA decorrelation do?
decorrelation + makes them statistically independent
What did Bell & Sejnowski (1995) propose about ICA and natural images?
They argued that natural images are generated by independent causes, and the brain may learn to separate these using ICA.→ Their model extracted components with high kurtosis (non-Gaussian), resembling receptive fields in visual cortex.
What is population coding in the visual system?
It’s when groups of neurons work together to represent a visual feature.→ Each neuron responds broadly, but the combined pattern gives precise info (e.g. orientation or motion).
On which three things does coding precision depend on?
sampling density
tuning width
response varaibility
How can you decode information from a neural population code?
By analyzing the pattern of activity across neurons using methods like vector averaging, maximum likelihood, or Bayesian decoding — to infer the most likely stimulus.
Name three influences/illusions you know for orinetation perception
zöllner illusion (Small oblique lines distort the perception of parallel lines, due to misinterpreted orientation signals in the visual cortex)
poggendorff illusion
tilt illusion
how are interaction btw neurons in a population coding model resembled?
they result in contextual effects similar to those found in perceptural experiments
What happens when color tuning curves overlap strongly in population coding?
Individual neuron responses become ambiguous, but the brain can still decode color from the overall population pattern, even under contextual influences.
Why do color-sensitive neurons have overlapping tuning curves?
Each neuron responds strongest to one color, but also weakly to similar colors,→ leading to overlapping tuning curves that enable fine-grained, population-based color representation.
How much of the cerebral cortex is concerned with the processing of visual information?
30-50%
What is the function of dorsal stream and ventral stream for vision?
d: processing of visual space, motion
v: processing of object information
they are not strictly sperated
What does remapping mean in dorsal stream when influencey by eye movements?
shortly before a saccadic eye movement, neurons become selective to stimuli at their postsaccadic. receptive field position
How are eye movements related to the dorsal visual stream?
The dorsal stream helps plan and update eye movements, encoding spatial targets and remapping object positions after saccades for stable visual perception.
Where does the coding of form and objects happen in ventral stream
V4
What are the two main visual processing streams and what do they do?
Ventral stream ("What"): object and identity recognition
Dorsal stream ("Where/How"): spatial processing and action guidance
Where does the coding of faces take place in the ventral stream
TE and STS
Last changed3 days ago