Lecture 7 08-12-2020
Brain mechanisms of learning and memory
What is happening in the brain during learning and memory formation?
Where does memory occur? First you have to answer that question.
An important concept in cognitive neuroscience is that of representations. Somehow the world is represented
in the brain. The defining function of nervous systems is representational.
Chruchland & Sejnowski:
So, what nervous systems do, they represent certain states of the world outside the brain.
Brain states represent states of some other system the outside world or the body itself
Representations that are happening now are patterns of activation across the units in a neural net (neurons).
They deal with incoming information by being activated.
Stored representations, by contrast, are believed to depend on the configuration of weights between units
crucial concept in CNW and learning memory.
In neural terms: these weights are the strength of synaptic connections between neurons. The sum is how
information is represented in the brain.
What is happening in the brain?
Tanzi & Hebb: an alteration in the effectiveness of existing connections. We do not need new connections.
most people in the field agree with this view.
Cajal: formation of new connections between neurons (synaptic).
Both would agree that during learning and memory formation, structural changes in neuronal connectivity, at
the level of the synapse, are happening changes in how they are connected.
Where do the changes occur?
Lashley: he was quite disappointed at the end of his career, he sometimes felt that learning just is not possible,
because he could not find a trace of it in the brain. He did found that, the bigger he made the lesion, the bigger
the effect on learning and memory, but that did not depend on the location.
Human amnesia
Amnesia = memory loss. Clinical cases of amnesia were studied.
Severe amnesia Clive Wearing: herpes simplex virus encephalitis leading to severe medial temporal lobe
damage. He does know general facts and can still play the piano. Apart from that, most what he is being told,
he forgets immediately.
Patient HM suffered from a serious form of epilepsy. He was operated in 1957, when parts of his temporal
lobes were removed. The surgeon thought he cut out part of the hippocampus, amygdala, and the cortex. But
the focus very soon was on the hippocampus (in MTL). That was the main damage to the brain.
Brenda Milner tested HM in detail. It became clear that the operation was successful in terms of epilepsy, but
there was a big effect on his memory. He became severely amnesiac.
Whatever the causes are of amnesia; some memory functions are intact.
Declarative memory is always impaired in these patients specific events
Procedural (nondeclarative) memory is often intact skills, simple forms of classical conditioning, priming
Non-delarative (implicit memory)
Claparède’s drawing pin anecdote that might make it clearer. He put a drawing pin (punaise) in his hand
when he shook hands with his patients. The patients got stabbed. The next week, the patient would come back
and they would not shake Claparède’s hand. The patient did not know why, but did not want to shake hands.
The remembered, but without awareness implicit, and it can’t be made explicit.
Skill learning mirror-drawing task. They have to do this by watching it through a mirror. Normal subjects take
some time to get better at it, but HM got better as well HM the next day, he started the task with the
improvement made last day. He clearly learned and the next day he remembered how to do the task. However,
,when HM was asked about this task, he denied any knowledge what he had just been doing. He had this
implicit memory of the task, but no explicit memory whatsoever.
Amnesiac patients and control subjects also perform the same on a mirror reading task. If you repeat the words
read, you have already seen it before, so you are quicker at identifying that word. An amnesiac (NA) did not
have the memory, but the skills were there to perform the task. He could just not benefit form episodic
memory of having seen specific words before.
Priming task subjects are given a list of words. To help recollection of these words, the first three letters are
given (priming).
Three kinds of instructions:
- Free recall which words can you remember
- Cued recall which words can you remember that start with these letters?
- Completion fill in the word that first comes to mind no reference is made to memory
The amnesiacs are not doing much words on the cued recall task. With the completion task, amnesiacs are just
as good. Clearly, there is memory there, but they are not aware of this.
Classical conditioning CS (bell, neutral) US (food, motivationally relevant) pairing
Before conditioning, the animal responds to the US, but not to the CS. At this stage, this is called the UCR.
During the conditioning the CS and the US are presented paired. The conditioning leads to a CR behavior
towards the formerly neutral stimulus.
Acquisition and extinction of an eyeblink. Tone = CS, followed by air being blown on eyeball of rabbit = US
UCR = blink. After a few pairings, the CR = blink when the tone is heard. If the CS is being presented without the
US, the rabbit stops responding after a few trials, no more CR, because the rabbit has learned that the CS is not
followed by the US extinction.
This kind of conditioning was done with HM. Like the rabbit, HM could be successfully conditioned. After the
experiment had finished, Milner would sit down with HM, and HM denied having done this experiment. Yet
again, you can see there is no explicit memory of what happened to him, but there is implicit memory of the
task.
Eyeblink conditioning: cerebellar lesions
This was unimpaired with HM lesion MTL.
Eyeblink conditioning is impaired by lesions to the cerebellum. But it is not that simple that this kind of
conditioning involves procedural memory instead of declarative memory.
There are two varieties of this kind of conditioning.
Delay the presentation of the CS and the US overlap.
Trace interval between end of CS and beginning of US.
Trace conditioning is more difficult. In HM, delay conditioning was done.
Delay vs trace conditioning in amnesiac patients:
Only the control subjects that were aware of this contingency (toevalligheid) between CS and US (trace
conditioning) (so when they were asked afterwards, the patients were aware of this). They could be
conditioned, but the ones unaware (control and amnesiac) could not be conditioned. Trace conditioning is
quite different form delay conditioning. Delay conditioning could be called to involve procedural memory (not
impaired in amnesia), whereas trace conditioning involves declarative memory awareness. And if there is no
awareness, there is no conditioning.
,Unlike delay conditioning, trace conditioning involves declarative memory. Declarative memory is impaired
when there are lesions to the MTL.
Forms of amnesia
Anterograde amnesia amnesia for events after surgery/trauma
Retrograde amnesia amnesia for events before surgery/trauma
Retrograde amnesia
Controls quite good memory of recent events in the years before testing, slightly declines when you go
further back in time
Korsakoff’s bad memory for recent events, but the further you go back in time, the better the memory
becomes
Alzheimer’s some curve/retrograde amnesia, but not as good as in Korsakoff’s.
Temporal lobe amnesia
Which brain structures are involved, hippocampus or extra-hippocampal cortex?
This was tested in an experiment involving a delayed-nonmatch-to-sample (DNMTS) task in monkeys.
Control group do very well on this task
Lesion restricted to hippocampus (H) doing quite well as well
Lesion to hippocampus and some lesions to overlying cortex (H+) performing quite badly, significantly worse
than controls an H subjects.
Lesions to hippocampus and more lesions to overlying cortex (H++) performing very bad, the worst in terms
of performance.
These results suggest that the overlying cortex is important for this particular memory task.
Later MRI of patient HM gives a different picture:
The MRI showed extensive damage to the medial temporal lobe. The damage they found was to the
hippocampus, but there were also parts of the hippocampus undamaged. On the other hand, there is quite a
lot of damage to the entorhinal cortex and the perirhinal cortex.
Varieties and mechanisms
Neural plasticity: when do changes occur?
Hebb: simultaneous activity in two neurons. Memory changes the structural connectivity. These changes occur
when there is simultaneous activity in two neurons = Hebb’s principle/Hebb’s synapse.
Cells that fire together, wire together (this is not a good answer on an exam, because it does not really explain
what is going on. This is better: changes in the effectiveness of synaptic transmission take place as a result of
simultaneous pre- (incoming action potential) and postsynaptic activity (depolarization) key principle in the
study of learning and memory.
Neurobiology of learning and memory: two research strategies, top-down and bottom-up
- Top-down: presupposing/assuming a certain principle (underlying memory)
o E.g.: LTP and spatial memory. LTP = long term potentiation, discovered by Lømo and Bliss.
Hippocampus is a bilateral structure, and has a temporal and septal pole. Information from
the entorhinal cortex enters the hippocampus in the dentate gyrus. From there, the cells the
synapse onto other cells in a part called CA3. They in turn synapse onto CA1 cells. This is a
trisynaptic circuit. Finally, the fibres leave the hippocampus again.
Bliss and Lømo put electrodes in the fibers from the entorhinal going in to the hippocampus
(stimulating) and in the dentate gyrus (recording electrode). They measured the response of a
group of neurons in the hippocampus.
LTP single stimulation of perforant path fibers to dentate gyrus results in an EPSP
(excitatory postsynaptic potential). In this case, because you measure a group of neurons, it is
a field potential. After a brief tetanus (250 Hz) (high frequent pulses, instead of one pulse) the
characteristics of the EPSP have changed. Then they resumed normal practice. The response
to a single pulse had changed as result of the tetanus. Two parameters changed:
Slope steeper
Amplitude of the population spike (measure of number of action potentials in the electrode
area, the more cells are firing, the greater this spike is) bigger
This was a relatively long-term change in responsiveness to a single electric pulse. So, the
characteristics of these neurons have changed. After the tetanus, every time you drive a little
, pulse in the hippocampus, there is a bigger response than before. This lasts for several hours.
After several days, the amplitude is still significantly increased. The response from cell B to
input form cell A has increased.
Here we have a long-term structural change in the characteristics of neurons in part of the brain often
implicated in memory. This might be some kind of mechanism that is also underlying learning and memory
long term changes in characteristics/connectivity of neurons.
Induction of LTP is dependent on a certain type of glutamate receptors, N-methyl-D-aspartate (NMDA)
receptors (protein in cell membrane).
The neurotransmitters from incoming cells can bind to the NMDA receptor. Normally, the channel opens when
the nt binds. Then, ions can flow in and out of the cell. But this is a particular kind of receptor, because not only
do we need binding of the nt to the receptor site, we also need to remove a block of the channel (magnesium).
It can be removed by depolarizing the cell (B) (incoming signal from cell A). So only when glutamate is binded
and the cell depolarizes at the same time, something happens, the calcium ions can go into the cell. This leads
to enhanced synaptic transmission after the induction of LTP. The calcium ions that could go into the cell cause
all kinds of second messenger activity (protein synthesis, and more glutamate receptors are being formed). So,
there are more glutamate receptors after LTP. This is a possible mechanism how LTP leads to structural changes
in these neurons more postsynaptic receptors. This is why neurotransmission is strengthened/increased.
So, it requires two simultaneous events: depolarization and glutamate in the cleft. This is exactly what Hebb
said.
NMDA, LTP, and memory
There are specific competitive blockers of NMDA receptors, e.g., AP5. These compete with the nt’s so that the
LTP can’t be induced. In this way, the effects of NMDA receptor blockade on LTP and on memory can be
studied, if it is the case that this is the same mechanism that underlies memory.
LTP and memory investigated by Richard Morris in a water maze (measures spatial memory).
Rats had to find a platform in this water maze, so they do not have to swim. When the rat found the platform,
it also looks at the environment. The rat was put in the pool at different positions, but the location of the
platform did not change. These rats can learn this very well. Clearly, they look at the environment. Apparently,
the rat can memorize the spatial layout of the room and during the next trial, it tends to swim to the platform
much more quickly. It seems to orient itself on spatial cues around the pool to find the platform.
Transfer test platform is moved, rat hovers around where the platform used to be.
Morris put electrodes in the hippocampi of these animals and did measurements. He also had a stimulating
electrode that would induce LTP. After LTP induction, the slopes have gone up significantly and the spike is
bigger. Another rat had a blocker infused with NMDA receptors (DL-AP5). Small quantities of this blocking
substances are being infused into the brain. If you try to induce LTP, it simply does not work. There is no long
term potentiation going on when you block NMDA receptors.
So, the induction of LTP can be successfully blocked, but what happens to memory the rats did not spend
much more time in the quadrant where the platform used to be compared to the other quadrants. It does not
seem to have any knowledge about where the platform used to be.
We can block LTP and at the same time we can block memory. The hypothesis that the mechanism underlying
LTP is the same as underlying memory has been confirmed by these results.
- Bottom-up: no presuppositions about the mechanism, but attempts at localization and what is
happening during memory formation
o E.g.: imprinting
Filial imprinting the formation - through learning – of an early social preference for the
mother or another stimulus they first encounter.
Konrad Lorenz was the first to describe this process. If the mother goose was removed and
replaced by himself, the gosling would treat him as the mother (follow him around etc.).
Artificial objects also sufficed as stimuli. You can train these animals to develop a preference
for these objects (spending time with them). So it is a very powerful learning mechanism.
