Introductory lecture
Vision is the primary sense, almost every action we make is guided by vision and a huge
proportion of the brain is devoted to visual processing. It is easy to study because:
● We can give visual inputs easily
● We can see where the eye is looking, so how the retina is being stimulated
● We can test what a person can perceive with behavioral methods
● We can access the visual cortex easily
Aspects of vision such as form, motion, colour, depth and distance, and attention and awareness
begin in the eye and share early processing stages. Later they are separated into different
functions performed by different pathways.
In early vision, we can see how individual neurons are processing information to help us
understand the sensory inputs from the world. Neural computation: an individual neuron
responds only to specific orientations of light.
Monism → mind is an aspect of the body and body is included in our mind via the nervous
system (mind and body are manifestations of the same physical thing).
Dualism → mind and body are separate.
Sensation → translation of external physical environment into a pattern of neural activity by a
sensory organ (monism view).
Perception → analysis of this neural activity to understand the environment and guide
behaviour, or the subjective conscious experience of the outside world (more broad than
sensation). Sensation and perception depend on physical properties of the world and are limited
by the physical properties of our sensors (for example sound and vision have different properties
and wavelengths/travel distances).
Perception has undergone evolution which helps us survive and reproduce:
● Optimized for useful representations of the environment
● Influenced by interpretation
● Dependent on limited resources of attention and awareness
Perception is often inaccurate since it is strongly influenced by experience and expectations, and
because our sensory organs receive more input than we can process.
The old way to study perception is psychophysics: testing whether subjects perceive a
difference between two stimuli. Just noticeable difference → the range of changes in the
stimulus that we can only just perceive, so when a difference is noticed. This depends on the
stimulus that is being compared.
The perceived intensity of a stimulus doesn’t increase with its physical intensity, instead we
perceive intensity in units of doublings (orders of magnitude) → logarithmic relationship.
Method of instant stimuli → subjects are presented with different extents of stimuli differences
along trials, and the detection probability is compared with the difference extents. This results in
a psychometric function/sigmoid-curve, where the steepest point is where the smallest
stimulus difference increases the perception the most. Adaptive staircases include an estimate
of the threshold and focuses on the measurements on that stimulus difference instead of the
whole psychometric function.
Nowadays, to study perception a measure of neural activity (spiking, synaptic, metabolic activity)
is correlated with a change in the presented stimulus or the behaviour.
● Spikes/APs of individual neurons are the clearest measure, very small and must be
measured directly from the neuron (invasive recordings). V1 neurons don’t respond to
, light spots (such as in the retina), but to more complex patterns emerging from spatial
relationships between multiple inputs of ganglion cells. V1 neurons can be orientation
specific → strongest response when a line/edge is in a specific direction.
● Synaptic activity/LFPs of groups of neurons can be measured with EEG
(non-invasive). This activity occurs in oscillations of different frequencies (theta-sleep,
alpha-inhibition, gamma-excitation), this pattern of excitation and inhibition also affects
perception → just noticeable differences are more likely to be perceived at an oscillation
peak.
● Metabolic activity/BOLD-signal can be measured with fMRI (non-invasive). It
measures the changes in blood flow and oxygenation, so how tissue magnetic
interactions change over time. The spatial resolution is very high, but the temporal
resolution is a bit poorer, not because of the images but because blood
flow/oxygenation takes 10 seconds. Because EEG has a good temporal resolution, fMRI
and EEG are often combined.
Steps of MRI:
● Place the brain in a strong magnetic field so that the randomly oriented atoms become
aligned.
● Add the gradient / input RF (a weaker, changing magnetic field), putting the atoms out
of alignment.
● Remove the gradient / input RF, putting the atoms back into alignment. The emitting
energy exit RF is captured by receive coil sensors.
● The emitted energy is measured and reconstructed into an image. This T1 image
(structural MRI) shows the amount of energy released at each location.
● But since the atoms wobble around when they go back into alignment, they release
more energy before their final stopping. This wobble is sensitive to the interactions
between atoms, and when we measure just the wobble energy, we can construct a T2
image.
● When we measure changes/distortions produced by magnetically active substances that
give information about the amount of (de)oxygenation, we can construct a T2* image
(functional MRI).
Between T2 and T2* (fMRI signal) there can be signal loss when the blood is deoxygenated
(iron atom in hemoglobin is not bound to oxygen), because it strongly affects the T2* wobble.
However, what is measured for oxygen loss is the BOLD-signal (blood oxygenation level
dependent).
At t0 (rest) → low blood flow → some
oxygenated blood cells and some
deoxygenated.
At t1 (neural activity) → oxygen is used →
reducing oxygenated blood cells. But after
some time, there is a large increase in blood
flow to deliver more oxygenated blood cells
(overcompensation by the hemodynamic
response function/HRF) → main signal in the
BOLD response.
, Lecture 1 - The Eye
Photoreceptors are not equally distributed across the retina:
● Cones in the fovea (middle of the retina) for colour vision
○ Fewer and larger cones further from the fovea
○ Thick cell body
● Rods in the periphery for low-light vision
○ Larger receptive fields further from the fovea
○ Thin cell body
In the optic disk the axons from ganglion cells leave the eye (forming the optic nerve), there
are no photoreceptors here. The axons of photoreceptors are in Henle’s fibre layer and project
onto bipolar cells and horizontal cells, of which the bipolar cells further project onto ganglion
cells.
Photoreceptors have an inner segment and outer segment, the membranes of the outer
segment contain transmembrane proteins for signaling. When light hits the photoreceptors:
● GPCR rhodopsin (opsin bound to 11-cis retinal) becomes active because of a
conformational change of 11-cis retinal into all-trans retinal
● G-protein transducin is recruited and reduces cGMP levels
● Less cGMP is bound to Na+ channels in the outer segment, closing them
● K+ channels in the inner segment remain open
● Hyperpolarization of the membrane because of K+ efflux and no Na+ influx
● Decreased glutamate release at the synaptic terminal
In dark conditions, cGMP levels are high, Na+ channels open and there is depolarization of the
membrane because of K+ efflux and Na+ influx, increasing neurotransmitter release.
There is 1 type of rod (medium-wavelength, not for colours) and 3 types of cones:
● 64% long wavelength/L cones (red) with small receptive fields and high resolution
● 32% medium wavelength/M cones (green)
● 2% short wavelength/S cones (blue) with large receptive fields and low resolution
Some cones are low in sensitivity so that our brain can compare the activity of cones and
construct colour. Cones are more sensitive to photopic vision (high-light, during day) while rods
are saturated. Rods are more sensitive to scotopic vision (low-light, during night). Mesopic
vision (intermediate-light) is perceived by both rods and cones. During dusk, blue and green
colours are perceived better, and red colours are perceived darker (rods are insensitive to long
wavelengths/red). This Purkinje-shift is the result of the reduced activity of cones and increased
activity of rods as it becomes darker. During dark adaptation, after 10 minutes there is the
rod-cone break where rods become more sensitive than cones and become more active.
Rods are more sensitive than cones because they take longer to regenerate. They have a longer
integration window allowing temporal summation of weak stimuli (the accumulation of light
intensity over a short period of time). Because of this, fewer photons per time unit are necessary
for a response. The disadvantage is that there is no differentiation between flashes within this
time period.
Most mammals have trichromatic vision (three types of cones). Dichromatic vision occurs in
colour blindness (when one type of cone is not functional):
● Protanopia (L-cone missing) → unable to perceive long wavelengths (red)
● Deuteranopia (M-cone missing) → unable to perceive medium wavelengths (green)
● Tritanopia (S-cone missing) → unable to perceive short wavelengths (blue)
Vision is the primary sense, almost every action we make is guided by vision and a huge
proportion of the brain is devoted to visual processing. It is easy to study because:
● We can give visual inputs easily
● We can see where the eye is looking, so how the retina is being stimulated
● We can test what a person can perceive with behavioral methods
● We can access the visual cortex easily
Aspects of vision such as form, motion, colour, depth and distance, and attention and awareness
begin in the eye and share early processing stages. Later they are separated into different
functions performed by different pathways.
In early vision, we can see how individual neurons are processing information to help us
understand the sensory inputs from the world. Neural computation: an individual neuron
responds only to specific orientations of light.
Monism → mind is an aspect of the body and body is included in our mind via the nervous
system (mind and body are manifestations of the same physical thing).
Dualism → mind and body are separate.
Sensation → translation of external physical environment into a pattern of neural activity by a
sensory organ (monism view).
Perception → analysis of this neural activity to understand the environment and guide
behaviour, or the subjective conscious experience of the outside world (more broad than
sensation). Sensation and perception depend on physical properties of the world and are limited
by the physical properties of our sensors (for example sound and vision have different properties
and wavelengths/travel distances).
Perception has undergone evolution which helps us survive and reproduce:
● Optimized for useful representations of the environment
● Influenced by interpretation
● Dependent on limited resources of attention and awareness
Perception is often inaccurate since it is strongly influenced by experience and expectations, and
because our sensory organs receive more input than we can process.
The old way to study perception is psychophysics: testing whether subjects perceive a
difference between two stimuli. Just noticeable difference → the range of changes in the
stimulus that we can only just perceive, so when a difference is noticed. This depends on the
stimulus that is being compared.
The perceived intensity of a stimulus doesn’t increase with its physical intensity, instead we
perceive intensity in units of doublings (orders of magnitude) → logarithmic relationship.
Method of instant stimuli → subjects are presented with different extents of stimuli differences
along trials, and the detection probability is compared with the difference extents. This results in
a psychometric function/sigmoid-curve, where the steepest point is where the smallest
stimulus difference increases the perception the most. Adaptive staircases include an estimate
of the threshold and focuses on the measurements on that stimulus difference instead of the
whole psychometric function.
Nowadays, to study perception a measure of neural activity (spiking, synaptic, metabolic activity)
is correlated with a change in the presented stimulus or the behaviour.
● Spikes/APs of individual neurons are the clearest measure, very small and must be
measured directly from the neuron (invasive recordings). V1 neurons don’t respond to
, light spots (such as in the retina), but to more complex patterns emerging from spatial
relationships between multiple inputs of ganglion cells. V1 neurons can be orientation
specific → strongest response when a line/edge is in a specific direction.
● Synaptic activity/LFPs of groups of neurons can be measured with EEG
(non-invasive). This activity occurs in oscillations of different frequencies (theta-sleep,
alpha-inhibition, gamma-excitation), this pattern of excitation and inhibition also affects
perception → just noticeable differences are more likely to be perceived at an oscillation
peak.
● Metabolic activity/BOLD-signal can be measured with fMRI (non-invasive). It
measures the changes in blood flow and oxygenation, so how tissue magnetic
interactions change over time. The spatial resolution is very high, but the temporal
resolution is a bit poorer, not because of the images but because blood
flow/oxygenation takes 10 seconds. Because EEG has a good temporal resolution, fMRI
and EEG are often combined.
Steps of MRI:
● Place the brain in a strong magnetic field so that the randomly oriented atoms become
aligned.
● Add the gradient / input RF (a weaker, changing magnetic field), putting the atoms out
of alignment.
● Remove the gradient / input RF, putting the atoms back into alignment. The emitting
energy exit RF is captured by receive coil sensors.
● The emitted energy is measured and reconstructed into an image. This T1 image
(structural MRI) shows the amount of energy released at each location.
● But since the atoms wobble around when they go back into alignment, they release
more energy before their final stopping. This wobble is sensitive to the interactions
between atoms, and when we measure just the wobble energy, we can construct a T2
image.
● When we measure changes/distortions produced by magnetically active substances that
give information about the amount of (de)oxygenation, we can construct a T2* image
(functional MRI).
Between T2 and T2* (fMRI signal) there can be signal loss when the blood is deoxygenated
(iron atom in hemoglobin is not bound to oxygen), because it strongly affects the T2* wobble.
However, what is measured for oxygen loss is the BOLD-signal (blood oxygenation level
dependent).
At t0 (rest) → low blood flow → some
oxygenated blood cells and some
deoxygenated.
At t1 (neural activity) → oxygen is used →
reducing oxygenated blood cells. But after
some time, there is a large increase in blood
flow to deliver more oxygenated blood cells
(overcompensation by the hemodynamic
response function/HRF) → main signal in the
BOLD response.
, Lecture 1 - The Eye
Photoreceptors are not equally distributed across the retina:
● Cones in the fovea (middle of the retina) for colour vision
○ Fewer and larger cones further from the fovea
○ Thick cell body
● Rods in the periphery for low-light vision
○ Larger receptive fields further from the fovea
○ Thin cell body
In the optic disk the axons from ganglion cells leave the eye (forming the optic nerve), there
are no photoreceptors here. The axons of photoreceptors are in Henle’s fibre layer and project
onto bipolar cells and horizontal cells, of which the bipolar cells further project onto ganglion
cells.
Photoreceptors have an inner segment and outer segment, the membranes of the outer
segment contain transmembrane proteins for signaling. When light hits the photoreceptors:
● GPCR rhodopsin (opsin bound to 11-cis retinal) becomes active because of a
conformational change of 11-cis retinal into all-trans retinal
● G-protein transducin is recruited and reduces cGMP levels
● Less cGMP is bound to Na+ channels in the outer segment, closing them
● K+ channels in the inner segment remain open
● Hyperpolarization of the membrane because of K+ efflux and no Na+ influx
● Decreased glutamate release at the synaptic terminal
In dark conditions, cGMP levels are high, Na+ channels open and there is depolarization of the
membrane because of K+ efflux and Na+ influx, increasing neurotransmitter release.
There is 1 type of rod (medium-wavelength, not for colours) and 3 types of cones:
● 64% long wavelength/L cones (red) with small receptive fields and high resolution
● 32% medium wavelength/M cones (green)
● 2% short wavelength/S cones (blue) with large receptive fields and low resolution
Some cones are low in sensitivity so that our brain can compare the activity of cones and
construct colour. Cones are more sensitive to photopic vision (high-light, during day) while rods
are saturated. Rods are more sensitive to scotopic vision (low-light, during night). Mesopic
vision (intermediate-light) is perceived by both rods and cones. During dusk, blue and green
colours are perceived better, and red colours are perceived darker (rods are insensitive to long
wavelengths/red). This Purkinje-shift is the result of the reduced activity of cones and increased
activity of rods as it becomes darker. During dark adaptation, after 10 minutes there is the
rod-cone break where rods become more sensitive than cones and become more active.
Rods are more sensitive than cones because they take longer to regenerate. They have a longer
integration window allowing temporal summation of weak stimuli (the accumulation of light
intensity over a short period of time). Because of this, fewer photons per time unit are necessary
for a response. The disadvantage is that there is no differentiation between flashes within this
time period.
Most mammals have trichromatic vision (three types of cones). Dichromatic vision occurs in
colour blindness (when one type of cone is not functional):
● Protanopia (L-cone missing) → unable to perceive long wavelengths (red)
● Deuteranopia (M-cone missing) → unable to perceive medium wavelengths (green)
● Tritanopia (S-cone missing) → unable to perceive short wavelengths (blue)
