In this part of the lecture, I want to move on from this early processing and

almost reflexive motor reactions to the type of processing mechanism that you

would associate with possibly High of a, higher level representations.

And specifically, I want to touch upon the case of the recording of surprising

events in the world. The identification of surprising events

in the world are extremely important. If we predict, if we know what's going to

happen then we know also how we, how we need to react, and everything is fine.

the interesting stuff. The, our behavior has to be guided by

mistakes in what we think about the world.

These are the times at which we need to reevaluate what the world is doing to us.

And so the coding of surprise is important.

And surprise can happen both by totally over stimulating things we never saw

before. if the spaceships of the aliens they're

coming down in the center of Jerusalem, or things that we already encountered but

that occur at times that we don't expect them to happen.

And then we will usually talk about deviance detection.

How we study the expectations in the brain?

So, there's a basic approach that is a, that can be, that is used in many

different studies. It is based on generating expectations.

So give a sequence of stimuli that makes you predict something about the future

and then at some point, you do something that violates this expectation and you

measure the activity of the brain. Is by measuring electrical activity or by

measuring blood flow, like in, like you do when you do function element I or any

other way of measuring brain activity and you see whether the brain activity

changed because of the regulation of expectations.

If brains react to the surprise, you conclude that they represented, also, the

expectations. Here's one type of experiment like this.

each circle here represent an auditory stimuli that was presented.

So there are beeps and boops and the beeps requires a lot, this light blue

colored circle. And the boops with the dark blue are

rare. In this case, the measured signals are EG

signals. They are measured from electrodes that

are put on the head of a human subject. And following the experiment, you take

the responses to the beeps and you average them together in order to get the

average response to the beeps. And you take the boops, you average the

response of the boops together to look at the response to the boops.

And the notion is that since the beeps occurs a lot, and the boops are rare,

every time a boop appear, you viol, violated expectations, because the

natural expectation to form is that the next stimulus is going to be the previous

one, and the previous one usually is the beep.

When you do that you find that the brain activity that is associated with the

[INAUDIBLE].

With the wave of sound, with the deviant is more negative, than the brain activity

that was listed by the standard. Okay this difference.

Between the response to the, the this extra negatively that comes with

responses to the deviate sound. Here you see the difference between the

two is called mismatch negativity. It has been described first by a Finnish

researcher Risto Naatanem in the late 70s.

And has been studied in great details because it is evoked by many, I mean

under many different conditions. And it's relatively early.

It occurs about 150 milliseconds, peaks about 150 milliseconds after seeing its

onset and 150 milliseconds is far before information seems to reach consciousness.

Consciousness would start at about here. So the MMN.

mismatch negativity represents a, a, an early activity.

And activity that goes in the brain that eventually will direct the conscious

introspective perseption of the sounds. But by itself it's it's not it, it occurs

before the, be, before consciousness of the, of the stimulus occurs.

Now, it turns out that singular activity can be evoked also with musical with

violation of musical expectations. very early electrical activity report.

The regulation of fingers' expectations, I will show you an example.

This is taken from this example is taken from a big work that was done in the land

of Stephen Kirsch. this is one of these one of one fourth of

it came out of this of these studies. And what Stefan Koelsch and his students

did here, was to study reactions to the end of musical eh, phrases.

The so called Cadences in material that we took from [UNKNOWN] written by

[UNKNOWN]. So here's one of these Cadences, [MUSIC].

Okay, this sounds like a good termination of a musical phrase.

in musical terms, what you have here is a transition from a chord that sits on the

fifth reprise dominant to a chord that sits on the first degree, the tonic.

But in fact, this is not what Bach wrote. What Bach wrote is slightly different,

subtly different. [MUSIC] 'Kay, the difference is in this

last chord. This last chord instead of, instead of

being a chord on the, on the first degree, it's a chord on the sixth degree.

And this is still Functionally, a tonic, it's the end of the musical phrase, but

it's not a stable ending. When you hear this, you want the phrase

to go on and continue. It turns out that almost all cadences in

Bach waltz are like this. And only very few of them [INAUDIBLE] if

you count slow to the this corpus of of a musical works, is like that.

Okay then its up to the, to the 6, to the 6 [INAUDIBLE].

So this represent a surprising lenient event.

And it's obvious that Bach was using it only for special purposes.

It was using it when he wanted the s, the musical phrase not to quit, to end, but

not to quite end. Now in the experiment, the the

researchers used a third type of ending. This is an ending that's completely off

[MUSIC] so this is really unexpected by the even non musical listeners would

presumably detect the fact that this cord is not, should not have been the.

With this type of music. Now, in this what you see here, I gained

the responses to the expected chord. So this is a response to an expected

chord And this is the response to the extremely unexpected cord, okay?

So you have a large difference between them, okay?

And this is again negativity and so you've got, in these plots negativity is

plotted upwards so this is the [UNKNOWN], we say [UNKNOWN].

Difference between the two responses in the continuous line.

It goes up, but it represents negativity, and this negativity is again early.

It's about 150 milliseconds after the code onset and it looks pretty similar to

mismatched negativity. It's not the, exactly the same thing as

mismatched negativity. but it's very similar.

This is the same thing, comparing the expected with the somewhat unexpected

code. Again, the response to the somewhat

unexpected code is slightly more negative, and if you look at the

difference, you get again negative difference wave pointing upward.

At the same, at the same time so even when you do this extremely slight

extremely subtle change in the chords, the, this cadence to the sixth instead,

instead of to the first degree the brain, your brain still react to it or at least

the brain of the subject in a eh, in, in this eh, in this eh, study.

So this early reaction to [UNKNOWN] to, to sounds occur even with with musical

material. And this may be, become relevant when

we'll talk about eh, emotions. So over the years, a number of groups

found similar phenomena similar to mismatch negativity, not identical, but

similar to mismatch negativity, even when recording from [INAUDIBLE] in the

auditory system of animals, and even in anesthetized animals.

What I show you here, is the membrane potential fluctuation of a neuron in

order to [UNKNOWN] of [UNKNOWN] rat. So you the [UNKNOWN] potential, it's

around minus 70 milli volts and you'll see the spikes, nice big spikes, a little

bit overshooting. The way that the the usually this type

and from time to time the sound presentations here.

And this sounds are pure tones and they come in their beeps and boops like in

this mismatch negatively experiment that I showed you at the begining of this

section. Blue represents beeps.

Stand out, the common sounds and the red represents books, the red sound.

So, blue is [UNKNOWN] common sound, red is the [UNKNOWN] sound.

In this case, you see the representation of the common sound, the following

presentation of [UNKNOWN] sound and then the next representation of, again, of a

common sound. The common sound at this location did not

evoke much activity. The other sound evoke a spike, evoked a

spike. And then this one might have evoked a

spike as well. So this is one of the presentation of the

real sound. Later in the experiment.

Again, a common sound, [UNKNOWN] sound, common sound.

Again, this common sound did not evoke an activity, this spike got evoked by things

that happened, the spontaneous ongoing stuff that happened just before it's

presentation. This sound presentation did not evoke any

activity by itself. But when the rare sounds came you have

here a blip. And this blip represents a response.

I will prove that you in a moment. And again the next sound.

The next common sound that work a spike in this case.

And here's a third example. No response to the common sounds.

A blip evoked by the last sound and again a spike evoked by the following by the

following common sound. So, if I just take the one common sound

and the following real sound and I just float.

All of these cases together on top of each other so there is a lot of [UNKNOWN]

there are 25 traces [UNKNOWN] on top of each other.

I can see that the, the this common sound which is after many common sounds did not

evoke much activity. But, the following [UNKNOWN] it

[INAUDIBLE] a nice [UNKNOWN] produces itself time after time, okay?

So, this is the type of a phenomena that they look pretty similar to mismatch

negativity and the, they're not quite mismatch negativity they're only similar

they're not identical. but responses turn outs to be a.

Very sensitive to the type of suprises you could think a sound sequence lights

contain. For example, when I play a common

[UNKNOWN] soound, I can make the sequence more or less regular.

keeping the overall probability of presentation of the sounds the same.

So, here you have an irregular sequence composed of two tones.

This is one tone, a low and this is a high tone.

The high tone is rare it occurs but in addition to being rare is also occurs at

random locations. I can arrange these tones so that I have

the same gain. A low tone which is common, and a high

tone which is rare, but now the high tone occurs periodically, exactly every fourth

location in the sequence, and so it is the high tone has the same probability.

Here in these two sequences, but it is much less surprising here.

It turns out that even that neurons in the auditory cortex of anesthetized rats

are sensitive even to this type of to this subtle difference in the

organization of tone sequence. So we call these sequences random, and we

call these sequences periodic. And in this study that was run by uh,

[UNKNOWN] in my lab they recorded responses to these types of sequences so

there were sequences in which the standard and deviant where a random

intermix and other sequences in which the stand and deviant.

Occurred with the same probabilities. But the deviant presentation was

periodic. In this illustration, [INAUDIBLE] exactly

every fifth stimulus in the sequence. And then we can compare the response to

this sound, okay? When it is in the periodic condition.

And so it's expected. You know exactly when it will happen, and

when not. And in this sequence when it is to some

extent unexpected, it's still expected but it's less expected than here.

And if you compare these responses to the same sound with the same probability you

see that the less expected sound, these give rise to larger responses than the

more expected sound. Okay, although they have, they share the

same probabilities. Just as long term structure of the

sequeces that differentiates between them.

So here we see sensitivities that are already on the border of being

interesting for music perception because this is, these 2 things can be thought of

as melodies. This is a regular bowing melody.

This is. Something that's very difficult to

follow. [UNKNOWN] melodies will follow somewhere

in between these two extremes, they will include some [UNKNOWN] and some

deviations from [UNKNOWN] and what we seee here, is that neurons in auditory

cortex of, even of a

[UNKNOWN].

Can discriminate between this type of regular and irregular melodies.

So, what I showed you is that muons in many stations of the auditory system are

sensitive to, to surprise. In fact, this type of sensitivity seems

to start even below the, the auditory cortex.

But it is certainly a present and strong in the sensitive at the, the level of

auditory cortex. And this suggests that the, the detection

of surprise is actually a major processing task of the auditory system,

and by implication of all, of many, of other sensory systems as well.

In fact, we can see suprise responses again, by doing many many different types

of exoeriments and as usual, there's a lot of important fascinating details.

And again I urge you to look at these type of studies and go into these

fascinating phenomena.

Higher Order Processes of Sensory Information

突触，神经元和大脑

与讲师见面