Foundations of cognitive neuroscience - II
LECTURE 1- 9/04
16-17 april cancelled
Premises:
Cognitive functions are the direct product of the activity of the brain, this doesn’t exclude that the other
organs communicate with it (even if it’s the very central part).
It is a bidirectional relationship, imposing reciprocal constraints in a form of circular causality.
Historical background: functionalism vs constructivism. These old pov are not available anymore. The
ultimate goal of any discipline interested in the human mind is to characterize the laws of conversion among
the different incapsulated level. In which level we can explain a certain phenomenon?
Human cognitive skills -> not present at birth, need formal and explicit teaching to be acquired. There are some
skills that satisfy only the rst condition (eg walking, language, basic tool use, dancing, singing).
So what are our focus object? Reading/writing, math/science, religion, philosophy etc.
They all require a speci c and explicit teaching to be acquired.
Difference in cultural repertoires -> Whiten et al. found out that some groups of chimpanzee developed some
group-speci c inventions that needed to be passed from a generation to the other through a teaching, they’re
group-speci c because absent in the other groups. However, it is clear that, compared to humans, there’s
not that level of massive creativity.
READING=Ability to translate arbitrary signs into elements of the spoken language.
It extends our memory capacity and allows us to extend in space and time information
among people.
MATHS= Ability to use arbitrary symbols to mentally represent quantities, objects,
relations, and to mentally transform them.
They are both very recent in the history of our species, and even today they’re not
universal.
A paradox-> we’re not born with a reading or math module in our brain, however
neuroimaging shows a huge consistency across subjects and culture about the supports of
mathematic and reading contents. It’s the same mechanism of genetic and predisposed
functions (eg homunculus, motor cortex). So to sum up, it seems that we’re predisposed
but we’re not.
A way to explain this paradox is the idea of cultural “recycling” of cortical maps -> redirection
of ancient neuronal circuits (made to support very basic functions) that are plastic enough
to support also novel informations or abilities.
Concept of cultural recycling: exaptation - cooptation - preadaptation -> they all convey
to the idea that there are some trades that by chance were selected under certain
evolutionary pressures, in course of the evolution it came out that they were able to
support new functions, different from the one that initially had the initial evolutionary
advantage.
Brain and culture - which link?
The hp of cultural recycling of cortical maps has some hp from which several predictions
can be derived
1. The human brain at birth is not a tabula rasa but made up by specialized and plastic
modular systems.
2. Our neural architecture is highly limited by its pre-determined functions
3. It is determined by genetic rules which however also include a certain degree of
plasticity (itself genetically determined!).
4. The new cultural acquisitions are possibily only if they exploit this dose of plasticity.
5. Every cultural object has become such (had passed through generations) inasmuch it
has found an adequate «neuronal niche» in our brain.
Two examples: spoken language and social skills
Up to 20 yrs ago, it was hard to believe that the brain was already organized at birth (ofc
the input that it receives are still so important).
1
fi fi fi
, There are some predisposed modules, even if the inputs that we receive are so important.
You expose adults and newborns to spoken language -> in newborns that area already
responds to that speci c stimulus (genetic predisposition for those circuits to elaborate
these kind of stimuli).
The other example is about social cognition -> a very strong effect is that even a couple of
days after birth, the child tends look more at a face compared to another nonsense image.
To tell if the newborn prefers more a certain gure, it’s suf cient to look at the direction
they’re looking on. Of course this phenomenon represents social advantage: for the baby ,
looking at the faces represents a way of bonding. With electroencephalograpy, it was
observed that also in 3 months newborn a right lateralized occipital area was particularly
sensitive to faces. The fact that the baby was 3 months could make us think that the
response is just due to the fact that faces is the most common stimulus that the baby sees
multiple times a day.
In order to overcome this doubt, it was done in cimec an experiment with 3 days old
newborns. In the paradigm they used an EEG combined with the observation using a
camera to see where the baby was looking at, the EEG portions in which the baby was not
staring at the gure were left out. The effect in the attention for the face still happens with
just a gure with 3 points (minimal geometric con guration that attracts the attention of a
baby), that represents the simpli ed version of a face the newborns stared at it for several
minutes. They used an oscillatory paradigm, oscillating images at a xed frequency ->
the stimulus is presented on and off with a xed frequency, the basic idea is that the neural
systems that contain neuron that react to the presented stimulus, will start oscillating in
synchrony with the stimulus. Therefore, it’s called frequency tagging -> after applying
Fourier transform, the data are analyzed on the basis of the frequency space. What they
predicted is that where the stimulus is presented, there will be a peak in the power
spectrum. We can observe a peak at 0.8 Hz, the peak is higher in upright than inverted
faces -> sign of face-speci c response, the brain responds more to upright gures. With
EEG, is possible to use a probabilistic software to reconstruct the regions in the cortex
responsible for the measured electricity on the scalp. It dif cult due to the trajectory of
electricity through the scalp, but with newborns the smirring of the signal is much reduced,
so source reconstruction for the baby is more reliable. They found that the right occipital
temporal/parietal are more active than the left. Based on these results, we can conclude
that even in 2 days old babies, they prefer faces but also the part that encode faces
is already active. We do know what the babies have been looking at in the rst weeks of
life, some researchers have put a cap with a camera and the result is: the’ve been looking
most of the time faces.
Start-up systems
= neurocognitive systems that support basic evolutionary skills that then were re-adapted
for other things. They are the foundational building blocks for the acquisition of culture
based skills
Starting from the start up systems, we can do some assumptions and predictions.
3 assumptions and 4 predictions of the neural recycling hypothesis
A1: the brain is organized in modules. As the result of natural evolution, the brain organization is
subject to strong anatomical constraints. This early organization of the brain in neuronal maps
which coding preferences are present since early in development (“precursor maps”) in uences
later cultural learning. This organization is universal.
P1: the brain regions involved in a given cultural function (“cultural maps”) should be consistent
across individuals and cultures, and they should be systematically related to precursor maps
coding for a similar evolutionary-relevant function.
A2: Cultural acquisitions need to nd their «neuronal niche», i.e., a series of cerebral circuits which
basic (innate?) functions are suf ciently close to the novel culturally acquired function and
suf ciently plastic to undergo a partial re-conversion to support the novel use.
2
fi fi fi fi fi fifi fi fi fi fi fifi fi fifi fl
,P2: the speed and ease with which novel cultural skills are acquired should be determined by the
complexity of the recycling required
A3: Because neural plasticity is limited, when cortical territories initially dedicated to evolutionary
ancient functions are recruited to support novel cultural functions, their preceding organization can
be partially overwritten.
P3: It should be possible to observe some speci c changes (gains or losses) in the perceptual /
cognitive skills originally supported by the recycled cortical territory.
P4: There should be important cross-cultural invariants in the implementations of the cultural
inventions, and they should be ultimately traced back to universal neural constraints.
Number as a (high level) percept
Cardinality is something that we perceive but is also something in a very abstract level.
It was found out that babies spontaneously match the number of sounds to the n of images
(the elements were moving a bit to capture the baby’s attention). They tended to look more
to the congruent matches (NB in developmental research is usually the opposite, but what
matters is that they make a difference) the differences were signi cant -> they mentally
extract and compare numerical quantities.
Not only this sense of number is present very early, but babies (9 months old in the
experiment) are also able to perform transformations over numerical numbers, so to
perform proto-arithmetics. McCrink et al simulated operational addition and subtractions ->
the looking time increases consistently in the case of trials with the wrong outcome (they
look longer at the impossible event). NB we could interpret this as a reaction for surprise,
while for the previous experiment it could be possible that the baby in case of matching
quantities “captures” the fact that there is something to learn there.
Talking about probability, Teglas et al. did an experiment with visual con gurations of
probabilities. The babies looked surprised when the outcome was a highly improbable one.
LECTURE 2 - 10/04
McComb et al.: In a completely ecological situation, lion can extract the numerosity
through the eyes and comparing numbers in order to decide if to attack or escape (adjust a
survival behavior).
Other experiments were done using the imprinting mechanism: Rugani et al used the
imprinting technique to imprint the newborn chicks to 5 kinder eggs, the chick followed it
according to the imprinting mechanism, it happened that the chick goes when the larger
part of the object is going.
Pica et al wanted to study this capacity in another type of population: indigenous tribe of
the Amazon, they don’t have a written system and a real vocabulary of numbers, they use
terms generally referred to number in a general way. The researcher presents different
numbers of seeds and ask which word they would use to describe the situation: the further
we go, the less precise they were (in particular, decrease > 4). The research question is: is
the sense of numerosity similar to the one that was discovered in Western cultures? In
order to study that -> presented a can with a bunch of seeds that go in a certain speed, a
can with some seeds going in, then a third set outside the can, question: are there more
seeds outside or inside? -> the more distant are the 2 sets in terms of number, the easier
to answer (curves similar to tribe and Western culture). So, the perception of numerosity is
kinda similar to the perception of sensory continua such as weight, light etc (but for
numbers there are not sensors to detect that), it obeys that same psico sical laws.
Numerosity perception is regulated by the Weber’s law, which applies to all
perceptual continua
Weber’s law=The minimum difference that can be perceived between two magnitudes is
proportioned to the intensity of the magnitudes themselves. It’s the same thing with light,
weight, etc. it’s about proportion/ratio that characterize the sets (10% vs 100% in the ex),
look at the demo
3
fi fi fifi
, -> From the demo: it’s intuitive which one contains more dots, however the difference
between the rst couple is the same of the one in the second couple, however the
amounts are different and this changes our perspective. They tested this principle in a lot
of experiments. Method of constant stimuli = gives st sample of a given numerosity but
different in other features, then a single test. The single test could be very small (half of the
given numerosity) or very big (twice). As we go far away to the reference number, people
are always better to detect the change. Plus, the data plotted in a logarithmic scale (-> not
considering scaling but ratio), the 2 curves become equal. To conclude, numerosity is:
1. Spontaneously extracted (no need of training/experience)
2. Independent from the sensory modality
3. Represented in an approximate and compressed way (according to Weber’s law)
4. Spontaneously combined in proto-arithmetical operations (comparison, addition,
subtraction).
Where in the brain is this kind of function is localized and which type of mechanism
underlies?
1993 -> they looked at all these effects eg ratio, magnitude etc and proposed a model that
proposed the idea of a presence of neurons with gaussian tuning functions with a xed
width on a logarithm scale (more overlapping with bigger numbers). Many years after in
Germany they recorded single neurons in macaque brain, with a delayed match to sample
task. The test could match or not in number the sample, the task was to press the button in
case of matching. There were neurons that responded speci cally to numbers. Prefrontal
and intraparietal cortex were the regions of these neurons.
Firing rate = how many spikes a neuron generates for a stimulus. To study a certain
phenomenon, is common to give stimuli whose characteristics are put in a continuum.
Doing these, we can build the function of spiking rate. Some neurons preferentially
responded to different numerosity. The response was gradually decreasing depending on
how far was the stimulus to the preferred one, the tuning function was a Gaussian.
Looking at the curves, we can also observe that the bigger are the numbers, the less is the
precision of encoding. Then, they looked at the latency (to see if it was a parallel encode
or not) -> they found out a rst activation in parietal and then a later one in prefrontal. In
fact, during also the delay, there was a consistent spiking activity.
Connection with behavioral results -> the monkey tends to do more errors in case of
bigger numbers.
In another experiment, instead of showing sets of objects they showed sequencies-> there
were number neurons also in this case (fe neurons that preferred the second stimulus in
the row).
Number neurons were found also in other species (crows and chicks).
Direct electophysiological intracranial recordings in epileptic patients showed interesting
results: in 2022, some fMRI data were collected regarding this activity. How to nd the
speci c activity for numerosity? Of course the subtraction method is not precise enough
(improbable that a voxel only activates for a certain numerosity) they used instead the
fMRI adaptation method -> recently replaced by others, it consists on the
elecrophysiological phenomenon of repetition-suppression (the neural activity decreases
if the stimulus is repeated). Going further, they hypothesized that in each voxel, different
population code different aspects of the stimulus. NB we cannot see this neural population,
no spatial resolution for them, to solve this problem we take advantage of the suppression
mechanism -> If you present a stimulus but you precede the stimulus with the very same
one, you’ll observe a suppressed response. This result should not be seen with very
different stimuli. Plus, further away are the second stimuli from the rst one, the less the
adaptation should be present. Eg I present 3 dots for multiple times and then 4, I present 3
multiple times and then 9 -> in the st situation the brain will recover less from the
adaptation -> Recovery proportional to the distance
4
fi fi fi fi fi fi fi fi fi
LECTURE 1- 9/04
16-17 april cancelled
Premises:
Cognitive functions are the direct product of the activity of the brain, this doesn’t exclude that the other
organs communicate with it (even if it’s the very central part).
It is a bidirectional relationship, imposing reciprocal constraints in a form of circular causality.
Historical background: functionalism vs constructivism. These old pov are not available anymore. The
ultimate goal of any discipline interested in the human mind is to characterize the laws of conversion among
the different incapsulated level. In which level we can explain a certain phenomenon?
Human cognitive skills -> not present at birth, need formal and explicit teaching to be acquired. There are some
skills that satisfy only the rst condition (eg walking, language, basic tool use, dancing, singing).
So what are our focus object? Reading/writing, math/science, religion, philosophy etc.
They all require a speci c and explicit teaching to be acquired.
Difference in cultural repertoires -> Whiten et al. found out that some groups of chimpanzee developed some
group-speci c inventions that needed to be passed from a generation to the other through a teaching, they’re
group-speci c because absent in the other groups. However, it is clear that, compared to humans, there’s
not that level of massive creativity.
READING=Ability to translate arbitrary signs into elements of the spoken language.
It extends our memory capacity and allows us to extend in space and time information
among people.
MATHS= Ability to use arbitrary symbols to mentally represent quantities, objects,
relations, and to mentally transform them.
They are both very recent in the history of our species, and even today they’re not
universal.
A paradox-> we’re not born with a reading or math module in our brain, however
neuroimaging shows a huge consistency across subjects and culture about the supports of
mathematic and reading contents. It’s the same mechanism of genetic and predisposed
functions (eg homunculus, motor cortex). So to sum up, it seems that we’re predisposed
but we’re not.
A way to explain this paradox is the idea of cultural “recycling” of cortical maps -> redirection
of ancient neuronal circuits (made to support very basic functions) that are plastic enough
to support also novel informations or abilities.
Concept of cultural recycling: exaptation - cooptation - preadaptation -> they all convey
to the idea that there are some trades that by chance were selected under certain
evolutionary pressures, in course of the evolution it came out that they were able to
support new functions, different from the one that initially had the initial evolutionary
advantage.
Brain and culture - which link?
The hp of cultural recycling of cortical maps has some hp from which several predictions
can be derived
1. The human brain at birth is not a tabula rasa but made up by specialized and plastic
modular systems.
2. Our neural architecture is highly limited by its pre-determined functions
3. It is determined by genetic rules which however also include a certain degree of
plasticity (itself genetically determined!).
4. The new cultural acquisitions are possibily only if they exploit this dose of plasticity.
5. Every cultural object has become such (had passed through generations) inasmuch it
has found an adequate «neuronal niche» in our brain.
Two examples: spoken language and social skills
Up to 20 yrs ago, it was hard to believe that the brain was already organized at birth (ofc
the input that it receives are still so important).
1
fi fi fi
, There are some predisposed modules, even if the inputs that we receive are so important.
You expose adults and newborns to spoken language -> in newborns that area already
responds to that speci c stimulus (genetic predisposition for those circuits to elaborate
these kind of stimuli).
The other example is about social cognition -> a very strong effect is that even a couple of
days after birth, the child tends look more at a face compared to another nonsense image.
To tell if the newborn prefers more a certain gure, it’s suf cient to look at the direction
they’re looking on. Of course this phenomenon represents social advantage: for the baby ,
looking at the faces represents a way of bonding. With electroencephalograpy, it was
observed that also in 3 months newborn a right lateralized occipital area was particularly
sensitive to faces. The fact that the baby was 3 months could make us think that the
response is just due to the fact that faces is the most common stimulus that the baby sees
multiple times a day.
In order to overcome this doubt, it was done in cimec an experiment with 3 days old
newborns. In the paradigm they used an EEG combined with the observation using a
camera to see where the baby was looking at, the EEG portions in which the baby was not
staring at the gure were left out. The effect in the attention for the face still happens with
just a gure with 3 points (minimal geometric con guration that attracts the attention of a
baby), that represents the simpli ed version of a face the newborns stared at it for several
minutes. They used an oscillatory paradigm, oscillating images at a xed frequency ->
the stimulus is presented on and off with a xed frequency, the basic idea is that the neural
systems that contain neuron that react to the presented stimulus, will start oscillating in
synchrony with the stimulus. Therefore, it’s called frequency tagging -> after applying
Fourier transform, the data are analyzed on the basis of the frequency space. What they
predicted is that where the stimulus is presented, there will be a peak in the power
spectrum. We can observe a peak at 0.8 Hz, the peak is higher in upright than inverted
faces -> sign of face-speci c response, the brain responds more to upright gures. With
EEG, is possible to use a probabilistic software to reconstruct the regions in the cortex
responsible for the measured electricity on the scalp. It dif cult due to the trajectory of
electricity through the scalp, but with newborns the smirring of the signal is much reduced,
so source reconstruction for the baby is more reliable. They found that the right occipital
temporal/parietal are more active than the left. Based on these results, we can conclude
that even in 2 days old babies, they prefer faces but also the part that encode faces
is already active. We do know what the babies have been looking at in the rst weeks of
life, some researchers have put a cap with a camera and the result is: the’ve been looking
most of the time faces.
Start-up systems
= neurocognitive systems that support basic evolutionary skills that then were re-adapted
for other things. They are the foundational building blocks for the acquisition of culture
based skills
Starting from the start up systems, we can do some assumptions and predictions.
3 assumptions and 4 predictions of the neural recycling hypothesis
A1: the brain is organized in modules. As the result of natural evolution, the brain organization is
subject to strong anatomical constraints. This early organization of the brain in neuronal maps
which coding preferences are present since early in development (“precursor maps”) in uences
later cultural learning. This organization is universal.
P1: the brain regions involved in a given cultural function (“cultural maps”) should be consistent
across individuals and cultures, and they should be systematically related to precursor maps
coding for a similar evolutionary-relevant function.
A2: Cultural acquisitions need to nd their «neuronal niche», i.e., a series of cerebral circuits which
basic (innate?) functions are suf ciently close to the novel culturally acquired function and
suf ciently plastic to undergo a partial re-conversion to support the novel use.
2
fi fi fi fi fi fifi fi fi fi fi fifi fi fifi fl
,P2: the speed and ease with which novel cultural skills are acquired should be determined by the
complexity of the recycling required
A3: Because neural plasticity is limited, when cortical territories initially dedicated to evolutionary
ancient functions are recruited to support novel cultural functions, their preceding organization can
be partially overwritten.
P3: It should be possible to observe some speci c changes (gains or losses) in the perceptual /
cognitive skills originally supported by the recycled cortical territory.
P4: There should be important cross-cultural invariants in the implementations of the cultural
inventions, and they should be ultimately traced back to universal neural constraints.
Number as a (high level) percept
Cardinality is something that we perceive but is also something in a very abstract level.
It was found out that babies spontaneously match the number of sounds to the n of images
(the elements were moving a bit to capture the baby’s attention). They tended to look more
to the congruent matches (NB in developmental research is usually the opposite, but what
matters is that they make a difference) the differences were signi cant -> they mentally
extract and compare numerical quantities.
Not only this sense of number is present very early, but babies (9 months old in the
experiment) are also able to perform transformations over numerical numbers, so to
perform proto-arithmetics. McCrink et al simulated operational addition and subtractions ->
the looking time increases consistently in the case of trials with the wrong outcome (they
look longer at the impossible event). NB we could interpret this as a reaction for surprise,
while for the previous experiment it could be possible that the baby in case of matching
quantities “captures” the fact that there is something to learn there.
Talking about probability, Teglas et al. did an experiment with visual con gurations of
probabilities. The babies looked surprised when the outcome was a highly improbable one.
LECTURE 2 - 10/04
McComb et al.: In a completely ecological situation, lion can extract the numerosity
through the eyes and comparing numbers in order to decide if to attack or escape (adjust a
survival behavior).
Other experiments were done using the imprinting mechanism: Rugani et al used the
imprinting technique to imprint the newborn chicks to 5 kinder eggs, the chick followed it
according to the imprinting mechanism, it happened that the chick goes when the larger
part of the object is going.
Pica et al wanted to study this capacity in another type of population: indigenous tribe of
the Amazon, they don’t have a written system and a real vocabulary of numbers, they use
terms generally referred to number in a general way. The researcher presents different
numbers of seeds and ask which word they would use to describe the situation: the further
we go, the less precise they were (in particular, decrease > 4). The research question is: is
the sense of numerosity similar to the one that was discovered in Western cultures? In
order to study that -> presented a can with a bunch of seeds that go in a certain speed, a
can with some seeds going in, then a third set outside the can, question: are there more
seeds outside or inside? -> the more distant are the 2 sets in terms of number, the easier
to answer (curves similar to tribe and Western culture). So, the perception of numerosity is
kinda similar to the perception of sensory continua such as weight, light etc (but for
numbers there are not sensors to detect that), it obeys that same psico sical laws.
Numerosity perception is regulated by the Weber’s law, which applies to all
perceptual continua
Weber’s law=The minimum difference that can be perceived between two magnitudes is
proportioned to the intensity of the magnitudes themselves. It’s the same thing with light,
weight, etc. it’s about proportion/ratio that characterize the sets (10% vs 100% in the ex),
look at the demo
3
fi fi fifi
, -> From the demo: it’s intuitive which one contains more dots, however the difference
between the rst couple is the same of the one in the second couple, however the
amounts are different and this changes our perspective. They tested this principle in a lot
of experiments. Method of constant stimuli = gives st sample of a given numerosity but
different in other features, then a single test. The single test could be very small (half of the
given numerosity) or very big (twice). As we go far away to the reference number, people
are always better to detect the change. Plus, the data plotted in a logarithmic scale (-> not
considering scaling but ratio), the 2 curves become equal. To conclude, numerosity is:
1. Spontaneously extracted (no need of training/experience)
2. Independent from the sensory modality
3. Represented in an approximate and compressed way (according to Weber’s law)
4. Spontaneously combined in proto-arithmetical operations (comparison, addition,
subtraction).
Where in the brain is this kind of function is localized and which type of mechanism
underlies?
1993 -> they looked at all these effects eg ratio, magnitude etc and proposed a model that
proposed the idea of a presence of neurons with gaussian tuning functions with a xed
width on a logarithm scale (more overlapping with bigger numbers). Many years after in
Germany they recorded single neurons in macaque brain, with a delayed match to sample
task. The test could match or not in number the sample, the task was to press the button in
case of matching. There were neurons that responded speci cally to numbers. Prefrontal
and intraparietal cortex were the regions of these neurons.
Firing rate = how many spikes a neuron generates for a stimulus. To study a certain
phenomenon, is common to give stimuli whose characteristics are put in a continuum.
Doing these, we can build the function of spiking rate. Some neurons preferentially
responded to different numerosity. The response was gradually decreasing depending on
how far was the stimulus to the preferred one, the tuning function was a Gaussian.
Looking at the curves, we can also observe that the bigger are the numbers, the less is the
precision of encoding. Then, they looked at the latency (to see if it was a parallel encode
or not) -> they found out a rst activation in parietal and then a later one in prefrontal. In
fact, during also the delay, there was a consistent spiking activity.
Connection with behavioral results -> the monkey tends to do more errors in case of
bigger numbers.
In another experiment, instead of showing sets of objects they showed sequencies-> there
were number neurons also in this case (fe neurons that preferred the second stimulus in
the row).
Number neurons were found also in other species (crows and chicks).
Direct electophysiological intracranial recordings in epileptic patients showed interesting
results: in 2022, some fMRI data were collected regarding this activity. How to nd the
speci c activity for numerosity? Of course the subtraction method is not precise enough
(improbable that a voxel only activates for a certain numerosity) they used instead the
fMRI adaptation method -> recently replaced by others, it consists on the
elecrophysiological phenomenon of repetition-suppression (the neural activity decreases
if the stimulus is repeated). Going further, they hypothesized that in each voxel, different
population code different aspects of the stimulus. NB we cannot see this neural population,
no spatial resolution for them, to solve this problem we take advantage of the suppression
mechanism -> If you present a stimulus but you precede the stimulus with the very same
one, you’ll observe a suppressed response. This result should not be seen with very
different stimuli. Plus, further away are the second stimuli from the rst one, the less the
adaptation should be present. Eg I present 3 dots for multiple times and then 4, I present 3
multiple times and then 9 -> in the st situation the brain will recover less from the
adaptation -> Recovery proportional to the distance
4
fi fi fi fi fi fi fi fi fi
