Theme Brain Region: Cerebral Hemispheres II: Multiple Cortical Maps.
Theme Animal: Pteronotus parnellii parnellii, the moustached bat, whose multiple cortical areas have assigned functions, unlike the numerous, functionally-unassigned multiple cortical visual maps in our own brains. An insight into the function of the cerebral cortex and its multiple maps has come from study of the bat's strange auditory world (which is like a movie of Fourier space in sound frequencies rather than our familiar movies of visual space). Such insight is presently impossible for our own visual cortex, despite superior human visual intuition and detailed knowledge of the location and connectivity of more than two dozen visual areas. We still do not know what all these areas do for us, in stark contrast to the clear functional significance of the bat's multiple representations of its auditory system.
Since mastering microbat sonar would take a semester or two in a class devoted entirely to it, I am going to take only one or two cortical areas of the microbat to make the point. You can take my word on the function of the microbat's other cortical areas. Nevertheless, if you have no background or interest in the physics of sound, this may be a little challenging for you.
In addition to their well-understood cortical function, microbats regularly measure microseconds in their daily lives (e.g. in the delay between emitting a pulse and hearing its echo, a tiny bat like Pteronotus can easily measure a distance of 1 mm, which corresponds to ~3 microseconds at the usual speed of sound in air...340 m/sec). Measuring microseconds is the second reason for making microbats the theme animal of this lecture.
Neuroethology:
This pursuit combines Ethology (the study of natural behaviour) with
Neuroscience (with diverse techniques for studying brain function). There
have been three Nobel Prizes for Ethology (von Frisch for decoding the
bee's dance, Tinbergen for innate releasing stimuli in birds and Lorenz
for imprinting) and too many to mention for Neuroscience (including the
Ozzie, Jack Eccles).......but none so far for the combination, neuroethology.
Last year's award to Kandel comes close, because he used Aplysia,
the sea slug, with its giant neurons as a model system to study the synaptic
bais of memory. Neuroscience meetings and neuroethology meetings are very
similar in the techniques used and the issues pursued, but there is a big
difference. ......meaningful behavioural context. At a neuroscience meeting
it is very likely that someone will have a spectacular picture of a new
finding, such as a monoclonal antibody that shows the specific distribution
of some new molecule in the brain. The problem is that such findings will
often be searching for a meaningful context. In contrast, neuroethologists
start from a favourite model system or organism that has a striking behaviour,
and they can usually explain the overall behavioural significance of their
findings, even the molecular ones.
I have peppered this module with a few examples from neuroethology and provided links so that you can follow up any of these rewarding studies that catches your fancy.
Declarative Memory and the Role of the Emotional landscape:
Exercise: Think back to your most visit memory (an exercise in declarative
memory that will search through the memory banks of your cerebral cortex).
I bet that your memory has a strong emotional component. Survival,
sex, hunger, thirst, fear, euphoria.......emotions whose activation has
two consequences:- 1. To activate neural systems (such as the diffuse
catecholanergic pathway) that "imprint" the ongoing sensory experience
by increasing the plasticity of the synapses, with the result that the
particular experience is laid down as a memory; 2. the collection of sensory
experiences associated with that event are now organised around their "emotional
landscape", which ties them together and gives them meaning.
The limbic system (medial cerebral structures including the cortical output from entorhinal cortex, cingulate cortex, amygdala, habenula and the brainstem monoaminergic nuclei). The limbic system can have focal damage that affects the "emotional landscape" without affecting perception, as in Capgras' Syndrome, where a small stroke, for example, has had the consequence that the emotional response to a loved one has been lost (as shown by an absent GSR (galvanic skin response) when the picture of the loved one is shown). Vision is completely intact, but in the absence of the meotional connection, the patient refuses to believe his/her eyes and decides that the loved one (sometimes even a pet) must be an imposter!
Capgras' Syndrome
Brilliantly described by Ramachandran, Capgras' syndrome is another
neurological syndrome that is often mistaken for insanity. The Capgras'
patient will typically identify people close to them as being imposters
- identical in every possible way, but identical replicas. Classically,
the patient will accept living with these imposters but will secretly "know"
that they are not the people they claim to be. When Capgras' occurs secondary
to other disorders such as the schizophrenias, the experience can take
on a frightening dimension along the lines of "Invasion of The Body snatchers".
Neurologically speaking, Capgras' Syndrome occurs when the kinesthetic
component of facial recognition is damaged. The result being that the strategy
of facial recognition no longer has the internal kinesthetic on which to
exit. Thus the patient gets caught up in an eternal loop in his strategy
and needs to form an explanation (the delusion) in order to understand
why this experience should occur.
Visit this great site with a large list of links on Capgras' Syndrome:http://members.spree.com/health/cotard/capgras.htm
Multiple Visual Cortical Maps: What are they for? This question
is a preoccupation of visual scientists who have mapped dozens of different
visual cortical areas, each repesenting the whole retina, in humans (using
fMRI) and animals. The somatosensory has at least 4 separate maps of the
body surface. Some functional differences have been found between different
cortical maps (there are separate visual areas for colour, motion and stereo,
for example). So it is a good bet that the different maps play functional
roles. The question remains:- What do they all do?
This question has been best answered in the cortex of the microbat,
Pteronotus
parnellii. Nobuo Suga has defined the function of a dozen or more different
areas in auditory cortex, each with a full tonotopic representation of
the cochlea, but each with different highly-specialised functions.
Despite the huge effort that has gone into defining the organisation of visual cortex, the functions of different areas are still not so clearly defined as they are in the case of the auditory cortex of this bat, whose task is so well-defined that it is possible to link different cortical functions to it. The functional organisation of visual cortex is very beautiful, but there is still not universal agreement about what it means.
click here for details of microbat echolocation in Pteronotus
Nobuo Suga:
Suga has spent his life studying microbat auditory cortex in behaving
bats. He gets invited to every international meeting on cortical organisation,
quite an achievement given the dominance of those working on visual cortex.
The cockpit of a bat:
There are a dozen different representations of the cochlea in Pteronotus
auditory cortex. Take just one example, the Combination-Sensitive region,
where the bat calculates the distance of the prey from the echo delay.
These neurons are so selective for a combination of auditory stimuli that
it would take a lifetime to discover the neuron's preferred stimulus ("trigger
feature") without prior knowledge of the particular sonar calls and echoes
to which the bat regularly listened. In fact, special oscillators had to
be constructed to deliver sounds with enough precision in the frequency
domain for these picky neurons. This may seem like unnecessary detail,
but it emphasises the point that we may have to know the precise behavioural
context before one can elucidate the function of a particular area.
Combination-Sensitive Area:
Here neurons are selective for the combination of one FM (frequency-modulated)
part of the call, followed with a delay, by the same FM component, or another
one (There are 3 harmonically-related FM components in each call....see
diagram). Unlike neurons in other parts of the bat's cortex, which respond
to parts of the call, such as CF1 or FM2, these neurons fire only if there
are two components separated by a precise delay. Different neurons have
different combinations and different delays which are mapped systematically
across the cortex.
In addition to revealing the way echo delay, and therefore prey distance, is mapped in the brain, study of this area helped reveal how the bat "knows" when to start counting the delay between the outgoing call and the returning echo.
Theoretically this could be by:-
We know that it must be self-stimulation, since the Combination-sensitive
neurons are of 3 types only, FM1-FM2, FM1-FM3, or FM1-FM4 (not any of the
other possible combinations such as FM2-FM3, FM3-FM1 etc etc).
We can make this inference because it is a common strategy for bats
to change the resonance properties of their vocal chambers to reduce the
amplitude of the first harmonic (CF1 and FM1). This is like putting on
a falsetto voice, which most of us can do at a pinch. With FM1 suppressed
by the bat, it may be just audible to the bat and neurons of this area
("self stimulation"), but too weak to generate an audible echo. This tends
to link the cortical processing strongly the to bat's own call (instead
of being susceptible to those from other bats flying around). In the specific
case of the combination-sensitive neurons, the suppressed fundamental is
being used to "initialise" the system so that echo delays can be measured.
Measuring the time of echoes
The FM-FM Combination sensitive neurons are arranged in an array of
systematically decreasing preferred delay in the cortical map as one goes
from back to front. Delays down to a few hundred microseconds can be measured
by these neurons. The longest delay coded for is around 18 msec, which
at the speed of sound in the tropics corresponds to ~600 cm, the
range within which the bat will "consider" the echo from a prey item. More
distant prey will not be detected by the cortex......hence the window of
acceptability. In a larger and faster bat, this window is set much larger.
General Lessons about the Cerebral Cortex that can be Learned from the Study of Bat Cortex:
1. Feature Filters:
Single neurons can have extraordinary specificity for a patten of stimulation.
In the case of the mustached bat, it would have taken an infinity of
time to discover the optimal stimulus for an FM-FM combination neuron had
the natural echolocation sounds not been available as a clue.
When Charlie Gross and his co-workers first decribed neurons
in monkey IT visual cortex that responded only to a face, or to a monkey's
hand, up to 9 hours was required to define the "trigger feature". Gross
was given a hard time by the neuroscientific establishment of the1970s
because such extreme specificity seemed unlikely ("yellow Volkswagen" feature
filters). Now the study of IT neurons in behaving monkeys is a cottage
industry, especially in Japan where creative teams of scientists work away
in IT.
2. Tunable Feature Filters:
The cerebral cortex is plastic, so the properties of the feature filters
can change if there is sufficient change in the inputs (e.g. amputation
experiments). In the case of the bat, we know that it grows up to respond
optimally to the frequencies it encounters, since males have deeper calls
with cortical neurons exactly matched in preferred frequency....... and
since rearing in helium will shift the preferred frequencies.
The ability of the cortex to adapt to the most commonly occurring features
helps to get around the "not enough neurons" problem. This problem is often
cited by those who reject the notion of feature filtering, on the grounds
that too many neurons would be required to cover all the possibilities.
This problem is moderated if the neurons do not have permanent assignment
to particular features, but are flexible (as we know they are from the
universal presence of LTP and all the experiments on cortical plasticity).
The flexibility probably exists in direct proportion to how far up
the hierarchy they are, with hippocampal neurons having the greatest plasticity
of all.
3. Multiple Cortical Maps:
The human brain probably has more than 3 dozen separate representations
of the complete visual field. What are they for?
Nearly everyone thinks that they may have separate functions (e.g. motion,
colour and stereo), but the details of those functions are pretty elusive.
In the case of the bat, we can have much confidence that each area has
a separate function, because we can relate the properties of the area so
directly to the function. For example, it is absolutely compelling that
the FM-FM combination area is concerned with echo ranging of prey, since
the properties of the area accurately predict the ranging behaviour of
the bat, including its 310 cm "window of acceptability". Moreover, the
details of the functional area illuminate mysterious aspects of the bat's
behaviour, a reverse flow of information that gives great confidence that
the functional significance of the areas we are attributing is correct.
For example, the properties of the FM-FM combination sensitive area make
it obvious that the bat is using self-stimulation with its own FM1 (First
harmonic) call to initialise the delay measuring process.
C. Other cases of microsecond timing:
i. Barn owl auditory localisation and the concept of a computational
map:
The barn owl has a computational map of sound source location that
has two axes involving two different stimulus parameters, neither of which
can be mapped directly from the cochlea. This is why it is a computational
map, rather than a conventioanl topographic map. But note that information
about sound source location is imparted to other brain regions in a topographic
way, even though the generation of the map is computational.
To generate the space map, information about two different sound modalities is first separated into two streams (intensity and time) so that small differences between the streams for each ear can be calculated. This gives two separate representations for the vertical dimension of sound space (because the ears are separated vertically to give intensity differences) and the horizontal dimension of sound space (because the time of arrival at each ear is different and can be used to give horizontal position). These two systems are then brought back together again to generate a map of both horizontal and vertical. This process of segregation, then fusion, is a common neural mechanism that enables the extraction and elaboration of time-related stimulus features into separate arrays when they might be confused if mixed before the abstraction took place. In the space map example, intensity changes cause alterations in the timing of impulses that could be confused with those produced by the time-of-arrival differences. Hence the separation of the two systems until horizontal and vertical have been effectively represented.
ii. Human auditory localisation:
Time of arrival differences at the two ears can be measured by humans
with an accuracy of 10-20 microseconds, just as in the barn owl.
Vertical is also judged by internsity differences, but in a more complicated
way that involves changes in the intensity of different parts of the sound
spectrum produced by the folds of the pinna and reflections off shoulders
etc.
D. Coincidence detection in a delay line: Using simultaneity to detect
non-simultaneous inputs.
By wiring up a neuron to both ears (binaural neuron), and varying the
length of the axons coming from each ear, one can set up a delay line.
The principle here is the same as the coinicidence detection already discussed,
except that coincidence on the neuron now means lack of coincidence in
the real world, on account of the conduction delays introduced by the wiring.
In this way, the brain can convert inter-aural time to place,
with adjacent neurons preferring delays that may differ by only 20 microseconds.
This explains the paradox of neurons using millisecond action potentials
to measure microsecond events. The cerebellum is time-to-place machine
par excellence, as already exxplained in Lecture 2 for the electric fish
JAR.