Temporal selectivity of identified auditory neurons in the cricket brain


Schildberger, K.

Journal of Comparative Physiology, A 155(2): 171-185

1984


The responses of 58 neurons in the cricket brain to auditory stimuli were studied by intracellular recording, after which the cells were marked with Lucifer Yellow. These auditory interneurons could be assigned to 3 classes on the basis of anatomical criteria: neurons (AN1 and AN2) ascending from the prothoracic ganglion into the brain, brain neurons (BNC 1) with arborizations overlapping those of ascending cells and brain neurons (BNC 2) with arborizations that had no overlap with those of ascending cells. All the ascending neurons had one projection field in common, an area in the arterior dorsal region of the diffuse neuropil lateral to the alpha-lobe. This field overlaps one of the projection fields of the BNC 1 neurons. Other projection fields of the BNC 1 neurons are frequently found in the posterior ventral region, at the boundary between the proto- and deutocerebrum; the latter overlap projection fields of the BNC 2 cells, the arborizations of which are often not clearly distributed in more than one distinct region. The threshold curves and suprathreshold responses revealed that all 3 classes include cells with low frequency (5 kHz) and others with broad-band (2-20 kHz) tuning; high-frequency (10-20 kHz) tuning was found only in AN and BNC 1 cells. Within the region to which the cells were tuned, sensitivity decreased in the order AN, BNC 1, BNC 2. A frequently observed property of BNC 2 neurons was independence of sound intensity in the suprathreshold intensity region. The response latency increased in the order AN, BNC 1, BNC 2. The effect of stimulus temporal structure was studied by presenting constant-energy-chirp trains of sound pulses (syllables) varying in duration and repetition interval. The low frequency AN1 neurons, like other prothoracic neurons, copy such signals over a broad range of repetition intervals. Copying by the BNC 1 neurons was less accurate; the responses of BNC 2 neurons were often not detectably synchronized with the syllables of the stimuli. Strikingly, the magnitude of the BNC 2 responses in some cases (7 out of 15) varied with the repetition interval; curves for the BNC 2 responses vs. repetition interval match closely those for the behavioral tracking response of females given an identical stimulus paradigm, and are independent of intensity at intensities 10 dB or more above threshold. The responses of AN and most BNC 1 neurons are uniform, regardless of stimulus intensity, over the entire range of repetition intervals studied. Some BNC 2 neurons responded to constant-energy chirps only with short and intermediate or with intermediate and long repetition intervals, resembling the high-pass and low-pass neurons found in frogs. This finding, the systematic latency differences and the anatomical relationships among the AN, BNC 1 and BNC 2 cells all point to a rather specific picture of the way that the male calling song is recognized.

Behavioral
Physiology
©
Springer-Verlag
1984
Physiology
A
J
Comp
Physiol
A
(1984)
155:171-185
Journal
of
Comparative
Neural
,
Temporal
selectivity
of
identified
auditory
neurons
in
the
cricket
brain
Klaus
Schildberger
Abteilung
Huber,
Max-Planck-Institut
fiir
Verhaltensphysiologie
D-8131
Seewiesen,
Federal
Republic
of
Germany
Accepted
May
25,
1984
Summary.
1.
The
responses
of
fifty-eight
neurons
in
the
cricket
brain
to
auditory
stimuli
were
studied
by
intracellular
recording,
after
which
the
cells
were
marked
with
Lucifer
Yellow.
These
auditory
inter-
neurons
could
be
assigned
to
three
classes
on
the
basis
of
anatomical
criteria:
(i)
neurons
(AN1
and
AN2)
ascending
from
the
prothoracic
ganglion
into
the
brain,
(ii)
brain
neurons
(BNC
1)
with
arboriza-
tions
overlapping
those
of
ascending
cells,
and
(iii)
brain
neurons
(BNC
2)
with
arborizations
that
had
no
overlap
with
those
of
ascending
cells.
2.
All
the
ascending
neurons
had
one
projection
field
in
common,
an
area
in
the
anterior
dorsal
region
of
the
diffuse
neuropil
lateral
to
the
alpha-
lobe.
This
field
overlaps one
of
the
projection
fields
of
the
BNC
1
neurons.
Other
projection
fields
of
the
BNC
1
neurons
are
frequently
found
in
the
posterior
ventral
region,
at
the
boundary
between
the
proto-
and
deutocerebrum;
the
latter
overlap
projection
fields
of
the
BNC
2
cells,
the
arboriza-
tions
of
which
are
often
not
clearly
distributed
in
more
than
one
distinct
region
(Figs.
1-4).
3.
The
threshold
curves
and
suprathreshold
re-
sponses
revealed
that
all
three
classes
(AN,
BNC
1
and
BNC
2)
include
cells
with
low
frequency
(5
kHz)
and
others
with
broad-band
(2-20
kHz)
tuning;
high-frequency
(10-20
kHz)
tuning
was
found
only
in
AN
and
BNC
1
cells.
Within
the
region
to
which
the
cells
were
tuned,
sensitivity
decreased
in
the
order
AN,
BNC
1,
BNC
2
(Figs.
2-4).
A
frequently
observed
property
of
BNC
2
neurons
was
independence
of
sound
inten-
sity
in
the
suprathreshold
intensity
region.
The
Abbreviations:
AN
Ascending
neuron;
BNC1
Brain
neuron
class
1;
BNC2
Brain
neuron
class
2;
BP
Band-pass;
HF
High-
frequency;
HP
High-pass;
LCT
Latency
corrected
timing;
LF
Low-frequency;
LP
Low-pass;
SC
Synchronization
coefficient;
SRI
Syllable
repetition
interval
response
latency
increased
in
the
order
AN,
BNC
1,
BNC
2
(Figs.
5,
6).
4.
The
effect
of
stimulus
temporal
structure
was
studied
by
presenting
'constant-energy-chirps',
trains
of
sound
pulses
(syllables)
varying
in
dura-
tion
and
repetition
interval.
The
low
frequency
AN1
neurons,
like
other
prothoracic
neurons,
copy
such
signals
over
a
broad
range
of
repetition
inter-
vals.
Copying
by
the
BNC
1
neurons
was
less
accu-
rate,
and
the
responses
of
BNC
2
neurons
were
often
not
detectably
synchronized
with
the
syllables
of
the
stimuli
(Figs.
7-10).
5.
Strikingly,
the
magnitude
of
the
BNC
2
re-
sponses
was
in
some
cases
(7
out
of
15)
found
to
vary
with
repetition
interval;
curves
for
the
BNC
2
responses
vs.
repetition
interval
match
closely
those
for
the
behavioral
tracking
response
of
females
given
an
identical
stimulus
paradigm,
and
are
inde-
pendent
of
intensity
at
intensities
10
dB
or
more
above
threshold.
By
contrast,
the
responses
of
AN
and
most
BNC
1
neurons
are
uniform,
regardless
of
stimulus
intensity,
over
the
entire
range
of
repeti-
tion
intervals
studied
(Fig.
11).
6.
Some
BNC
2
neurons
responded
to
constant-
energy
chirps
only
with
short
and
intermediate
or
with
intermediate
and
long
repetition
intervals,
resembling
the
high-pass
and
low-pass
neurons
found
in
frogs
(Fig.
12).
This
finding,
the
systema-
tic
latency
differences,
and
the
anatomical
rela-
tionships
among
the
AN,
BNC
1
and
BNC
2
cells
all
point
to
a
rather
specific
picture
of
the
way
that
the
male
calling
song
is
recognized.
Introduction
Female
crickets
are
attracted
by
the
species-specific
calling
song
of
their
males.
Two
parameters
of
the
172
K.
Schildberger
:
Auditory
neurons
in
the
cricket
brain
song
are
particularly
important
in
eliciting
and
maintaining
the
phonotactic
response
of
the
fema-
les:
spectral
composition
and
temporal
pattern
(Po-
pov
and
Shuvalov
1977;
Pollack
and
Hoy
1979;
Weber
et
al.
1981;
Thorson
et
al.
1982).
Whereas
the
auditory
system
of
the
cricket
exhibits
tuning
to
certain
frequencies
even
at
the
level
of
the
prima-
ry
nerve
fibers
(Nocke
1972;
Esch
et
al.
1980;
Hut-
chings
and
Lewis
1981),
no
specific
temporal
filter-
ing
is
discernible
at
this
level.
In
the
prothoracic
ganglion
further
processing
of
the
auditory
infor-
mation
from
a
number
of
local,
ascending
and
descending
neurons
occurs
(Casaday
and
Hoy
1977;
Popov
et
al.
1978;
Wohlers
and
Huber
1978,
1982;
Popov
and
Markovich
1982;
Boyd
et
al.
1984).
Binaural
processing,
which
could
be
used
for
directional
localization,
first
takes
place
at
this
level
(Kleindienst
et
al.
1981;
Wohlers
and
Huber
1982).
However,
neither
the
ascending
nor
the
descending
neurons
in
the
prothorax
are
specifically
tuned
to
phonotactically
effective
temporal
patterns
(Woh-
lers
and
Huber
1982).
These
neurons
copy
in
detail,
or
at
least
in
certain
parameters,
both
phonotacti-
cally
effective
and
ineffective
patterns.
In
the
brain,
too,
local
auditory
neurons
have
been
found
that
only
copy
the
temporal
structure
of
the
stimulus
pattern
(Boyan
1980).
Therefore,
a
main
problem
in
studying
the
audi-
tory
pathway
of
the
cricket
is
to
show
if
the
brain
contains
identifiable
neurons
with
specific
temporal
filter
properties
that
in
fact
match
the
behaviorally
demonstrated
recognition
parameters
of
the
con-
specific
song.
In
the
experiments
described
here,
both
the
anatomical
and
the
physiological
properties
of
in-
dividual
auditory
neurons
in
the
cricket
brain
were
examined.
The
frequency-,
intensity-
and
pattern-
dependence
of
the
response
of
the
brain
neurons
are
compared
with
those
of
two
ascending
neurons.
Materials
and
methods
Female
larvae
of
Gryllus
bimaculatus,
raised
in
the
laboratory,
were
isolated
before
the
last
molt
and
kept
in
isolation
until,
as
2-
to
3-week-old
adults,
they
were
used
in
the
experiments.
The
experiments
were
done
during
the
daytime
in
a
darkened
room,
at
room
temperature
(20-23°
C),
on
a
table
stabilized
and
shield-
ed
for
intracellular
recording.
After
removal
of
the
wings
the
animal
was
fixed
to
a
holder
by
its
back;
the
legs,
thorax
and
head
were
immobilized
and
the
mouthparts
and
gut
were
remo-
ved.
The
foreleg
tibiae
were
placed
in
miniature
sound
chambers
(`legphones';
Kleindienst
et
al.
1981).
The
frontal
surface
of
the
head
was
opened
between
the
lateral
ocelli,
the
medial
ocellar
nerve
was
transected,
and
tracheae
and
glands
overlying
the
brain
were
dissected
away.
A
metal
spoon
was
placed
beneath
the
brain
to
stabilize
it
and
to
serve
as
a
reference
electrode.
The
recording
electrodes
were
glass
micropipettes
(50-200
MS2)
fil-
led
with
3%
Lucifer
Yellow.
Intracellular
and
extracellular
potentials
were
amplified
and
recorded
in
the
conventional
man-
ner
and
analyzed
by
computer.
Whenever
the
intracellular
re-
cording
remained
stable
for
the
duration
of
a
15-min
stimulus
program,
the
cell
was
subsequently
injected
with
Lucifer
Yellow
by
a
hyperpolarizing
current
(3-10
nA,
5-60
min).
In
some
recordings
of
AN1
physiological
tests
were
also
done
during
injection
of
hyperpolarizing
current.
No
difference
in
the
re-
sponses
occurred
concerning
the
spontaneous
activity,
the
laten-
cy,
the
number
and
pattern
of
action
potentials,
compared
to
the
non-hyperpolarized
situation.
After
the
brain,
or
the
whole
ganglion chain
down
to
the
metathoracic
ganglion,
had
been
dissected
out,
the
ganglia
were
fixed
for
4-12
h
in
4%
aqueous
formaldehyde,
dehydrated,
cleared
and
photographed
as
a
whole
mount
in
the
fluorescence
microscope.
After
transfer
into
propylene
oxide
they
were
embedded
in
Spurr's
medium,
section-
ed
(15µm)
and
photographed
again.
The
neurons
were
recon-
structed
by
reference
to
both
whole
mount
and
serial-section
photographs.
The
auditory
stimuli
were
presented
by
way
of
legphones
(see
Kleindienst
et
al.
1981).
Tone
frequency
was
varied
between
2
and
20
kHz
in
1-
to
2-kHz
steps,
and
intensity
between
35
and
90
dB
SPL
in
5-dB
steps.
For
measurement
of
the
frequency-
and
intensity-dependence
of
the
neuronal
responses,
the
stimuli
consisted
of
'chirps'
140
ms
long
at
500
ms
intervals;
each
chirp
was
subdivided
into
4
20-ms
syllables
separated
by
20-ms
pau-
ses.
After
this
test
a
1-min
pause
was
made.
Then,
to
reveal
the
ways
in
which
temporally
structured
stimuli
are
represented
in
the
neuronal
discharge,
constant-energy
chirps
were
tested,
each
250
ms
long
and
comprising
syllables
with
various
repetition
intervals
(10-98
ms,
50%
duty
cycle
in
each
case)
and
constant
carrier
frequency
of
5
or
15
kHz.
To
test
for
hysteresis
the
chirps
were
presented
with
increasing
syllable
repetition
intervals
(SRI's)
from
chirp
to
chirp
over
the
range
of
10-98
ms
in
8
ms
steps,
and
then
with
decreasing
SRI's
from
98
to
10
ms;
the
chirp
repetition
interval
was
500
ms.
Whenever
possible
the
result
was
confirmed
by
a
second
test
one
minute
later.
Here
the
SRI
was
kept
constant
for
12
consecutive
chirps,
then
changed
unsystematically,
in
the
range
of
10-98
ms,
for
the
next
set
of
12
chirps
and
so
on.
The
patterns
in
these
to-and-fro
and
unsy-
stematic
paradigms
were
the
same
as
those
designed
to
characte-
rize
the
recognition
process
in
behavioral
experiments
on
pho-
notaxis
(Thorson
et
al.
1982).
When
a
single-frequency
carrier
is
amplitude
modulated,
the
frequency
spectrum
of
the
signal
is
altered.
This
small
alteration
must
be
tolerated
because
to
use
broad-band
noise
would
stimu-
late
high-frequency
receptors
and
pathways
not
activated
by
natural
calling
song.
These
could
complicate
central
responses,
for
they
are
involved
in
phenomena
such
as
two-tone
excitation
and
inhibition
(Boyd
et
al.
1984),
anomalous
phonotaxis
(Thor-
son
et
al.
1982)
and
possibly
bat-avoidance
reflexes
(Pollack
et
al.
1984).
Response
parameters
of
interest
were
the
latency,
the
num-
ber
and
the
temporal
pattern
of
the
action
potentials
in
the
discharge
occurring
during
the
300
ms
after
the
onset
of
the
chirp.
For
neurons
without
spontaneous
activity
the
threshold
stimulus
intensity
was
taken
to
be
the
intensity
that
elicited
at
least
two
action
potentials
per
chirp.
For
spontaneously
active
neurons
the
criterion
for
threshold
intensity
was
that
the
dis-
charge
rate
during
the
response
be
significantly
above
the
mean
rate
of
spontaneous
activity
(1-test,
p<
0.05).
The
pattern-cop-
ying
properties
of
the
neurons
were
analyzed
by
way
of
post-
stimulus-time
histograms
and
a
diagram
relating
the
neuronal
events
to
the
stimulus
pattern.
The
degree
of
synchronization
of
the
action
potentials
with
the
syllables
at
various
syllable
repetition
intervals
was
examin-
ed
by
constructing
a
latency-corrected-timing
(LCT)
plot
(see
Fig.
10).
Latency
here
is
defined
as
the
time
between
the
onset
K.
Schildberger
:
Auditory
neurons
in
the
cricket
brain
173
of
the
first
syllable
and
the
occurrence
of
the
first
action
poten-
tial
during
the
chirp.
'Latency
correction'
amounted
to
the
subtraction
of
this
time
from
the
actual
time
of
occurrence
(with
respect
to
the
stimulus
onset)
of
each
action
potential;
as
a
result
of
this
operation,
the
time
of
occurrence
of
the
first
action
potential
was
0
ms.
The
corrected
times
of
occurrence
of
the
remaining
action
potentials
associated
with
the
first
syllable
plus
the
first
action
potential
associated
with
the
next
syllable
were
plotted
(ordinate).
Then
a
similar
plot
for
the
second
syllable
was
superimposed
by
setting
the
onset
time
of
the
first
action
potential
equal
to
its
latency-corrected
time
minus
the
syllable
repetition
interval,
and
so
on
for
each
syllable
to
the
end
of
the
chirp.
This
procedure
was
repeated
for
each
of
the
syllable
repetition
intervals
tested
(abscissa),
to
produce
the
LCT
plot.
A
numerical
measure
of
this
kind
of
copying
capability,
the
synchronization
coefficient
(SC),
was
also
calculated;
after
la-
tency-correction
all
of
the
neuronal
events
associated
with
a
syllable
(ES)
and
those
occurring
during
a
pause
between
syllab-
les
(EP)
were
counted
and
the
ratio
(ES—EP)/(ES
+EP)
was
obtained.
In
the
case
of
perfect
copying
of
syllable
or
pause
the
SC is
1
or
—1,
respectively,
and
if
there
is
no
copying
at
all
SC
=
0.
The
significance
of
the
SC
values
so
obtained
was
evaluated
by
the
chi-square
test.
Results
Anatomy
Two
auditory
interneurons
ascending
from
the
prothoracic
ganglion
into
the
brain
were
identified
in
several
preparations
(Table
1);
each
of
these
was
stained
over
its
entire
course
from
the
prothoracic
ganglion
to
the
brain.
The
courses
of
the
axons,
positions
of
the
cell
bodies
and
dendritic
branching
in
the
prothoracic
ganglion
correspond
to
those
described
by
Wohlers
and
Huber
(1982),
from
pre-
parations
stained
in
the
prothoracic
region,
for
the
neurons
they
called
AN1
and
AN2
(AN
=
ascen-
ding
neuron).
Both
neurons
pass
through
the
sube-
sophageal
ganglion
with
little
or
no
arborization
and
end
in
the
protocerebrum.
The
AN1
axon
first
sends
out
branches
in
a
ventral
posterior
region
of
the
protocerebrum,
where
it
adjoins
the
deutocere-
brum;
the
axon
continues
forward
and
dorsalward
to
end
in
a
main
projection
field
lateral
to
the
a-lobe
(Figs.
1,
2).
AN2
passes
through
the
brain
in
a
similar
direction,
from
posterior
ventral
to
anterior
dorsal.
However,
the
path
of
this
axon
lies
some-
what
further
lateral,
curving
so
that
its
main
arbo-
rization
occurs
in
nearly
the
same
region
as
that
of
AN1.
Another
branch
of
the
AN2
axon
runs
in
a
posterior
dorsal
direction
to
the
point
of
entry
of
the
optic
nerve,
and
then
turns
anterior
and
arbori-
zes
at
the
uppermost
dorsal
margin
of
the
protoce-
rebrum
(Figs.
1,
2).
The
two
ascending
auditory
interneurons
found
so
far
project
into
a
relatively
narrowly
circumscri-
bed
region
of
the
diffuse
protocerebral
neuropil,
extending
along
an
anteroposterior
axis
at
a
depth
of
30-100
pm
and
measuring
ca.
200
pm
from
its
dorsal
to
its
ventral
margin
and
ca.
150-200
pm
from
the
edge
bounded
by
the
a-lobe
to
its
lateral
boundary.
In
addition
to
these
ascending
cells,
a
number
of
auditory
interneurons
were
found
that
lie
entire-
ly
within
the
brain
(Fig.
1,
Table
1).
The
following
classification
is
based
on
anatomical
characteriza-
tion
by
the
criteria
of
the
position
and
general
appearance
of
the
arborizations,
as
well
as
on
phy-
siological
characteristics
of
these
cells.
The
posi-
tions
of
the
cell
bodies
in
one
class
varied,
as
did
the
density
and
number
of
branches
in
the
individu-
al
projection
areas.
The
fundamental
criterion
for
classification
is
that
Brain
Neuron
Class
1
(BNC1)
comprises
cells
with
arborizations
in
the
main
projection
field
of
the
ascending
neurons,
whereas
Brain
Neurons
Class
2
(BNC2)
do
not
arborize
in
this
projection
field.
Only
those
neurons
were
called
'identified'
Table
1:
Characteristics
of
auditory
interneurons
in
the
cricket
brain
Anatomical
class
No.
of
stained
cells
Identified
No.
Name
Frequency
tuning
Synchronization
with
syllables
when
SRI
[ms]
Temporal
selectivity
in
response
magnitude
when
SRI
[ms]
Ascending
25
10
AN1
LF
>
=18
15
AN2
HF—BB
>
=34-58
Brain
neuron
class
1
18
4
BNC
la
2
BNC
lb
4
BNC
lc
LF
LF
HF
>
=
50
3
BNC
1
d
HF—BB
>
=
58
>26
(LP)
5
not
ident.
1LF,
4HF
>
=26,
58,
Brain
15
5
BNC
2a
LF
>26-58<
(BP)
neuron
2
BNC
2b
LF
<42
(HP)
class
2
2
BNC
2c
BB
>26-58<
(BP)
2
BNC
2d
BB
4
not
ident.
2LF,
2BB
174
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
BNC
2a
BNC
2c
BNC
2b
BNC
2d
BNC
1a
BNC
1c
BNC
1
b
BNC
1d
AN
1
AN
2
Fig.
1.
Identified
auditory
neurons
in
the
cricket
brain.
Recon-
structions
from
serial
sections
of
Lucifer
Yellow
stained
cells;
frontal
view;
low-frequency
neurons
are
shown
in
the
left
half
of
each
ganglion,
high-frequency
neurons
in
the
right
half;
dashed
lines
indicate
mushroom
bodies;
the
cells
have
no
direct
connec-
tions
to
the
mushroom
bodies
which
were
detected
by
anatomical
criteria
and
physiological
criteria
like
threshold
curve,
latency
and
response
pattern
as
the
same
cells
in
several
preparations.
These
identified
cells
are
shown
in
Fig.
1
and
some
of
their
physiological
characteris-
tics
are
listed
in
Table
1.
Among
the
BNC1
neurons,
some
arborize
only
within
this
field
(local
BNC1)
and
others
have
ex-
tensive
ipsilateral
branches
in
a
posterior
region,
between
the
deuto-
and
protocerebrum
(wide-field
BNC1).
Cells
in
the
latter
group
exhibited
no
thick-
ened
terminal
structures
in
the
arborization
region
overlapping
that
of
the
ascending
cells,
though
such
specializations
could
be
found
in
terminal
or
en
passant
positions
in
the
second
projection
field
of
these
neurons
(Figs.
1,
3).
BNC2
neurons
arbo-
rize
over
a
large
area
in
central
to
posterior
regions
between
the
deuto-
and
protocerebrum.
Here
they
overlap
arborizations
of
wide-field
BNC1
neurons.
BNC2
neurons
were
rarely
found
to
have
two
quite
Fig.
2.
Response
maps
of
the
ascending
neurons
AM
(top)
and
AN2
(bottom),
reconstructions
of
which
are
shown
on
the
right;
inset:
schematical
reconstruc-
tion
(not
to
scale)
from
side
view.
In
the
two
response
maps
the
responses
to
syn-
thesized
four-syllable
chirps
varying
in
intensity
and
carrier
frequency
are
plot-
ted;
response
magnitude
(total
number
of
action
potentials)
is
represented
by
bar
height,
and
dashes
indicate
no
re-
sponse.
Calibration
(for
both
graphs):
50
action
potentials.
Dashed
lines
mark
threshold
and
outline
the
inhibitory
side-
band.
The
same
kind
of
diagram
is
used
in
Figs.
3
and
4
90
-
70
I
I
I
I
a
I
I
I
1
I
.
1
1
.
.
50
-
30-
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
175
Intensity
90
-
70
50
30
I
5
10
20
2
5
10
20
Frequency
[kHz]
distinct
projection
fields
as
the
BNC1
neurons
do,
and
the
appearance
and
position
of
the
terminal
branches
provide
no
evidence
of
spatial
separation
of
the
main
input
and
output
regions
of
the
BNC2
neurons
(Figs.
2,
4).
There
are
thus
a
number
of
anatomical
indica-
tions
of
a
stepwise
transfer
of
auditory
information
in
the
brain,
from
the
ascending
neurons
via
the
BNC1
to
the
BNC2
cells.
Response
maps
for
standard-pattern
stimuli
For
comparison
of
the
effects
of
tone
frequency
and
intensity
on
the
various
types
of
neurons,
detailed
`response
maps'
were
constructed
by
stimulating
with
simulated
chirps
having
a
fixed
temporal
pat-
tern
(see
Materials
and
methods).
Typical
response
maps
are
shown
in
Figs.
2,
3
and
4.
Their
main
features
are
as
follows.
Thresholds
The
ascending
neurons
could
be
assigned
to
two
types
on
the
basis
of
their
threshold
curves:
low-
frequency
neurons,
which
corresponded
anatomi-
cally
to
AN1
(Fig.
2),
and
high-frequency
neurons,
corresponding
to
AN2
(Fig.
2).
AN1
neurons
were
most
sensitive
to
4-5
kHz
the
carrier
frequency
of
the
calling
song
at
35-40
dB
SPL,
though
for
occasional
AN1
neurons
the
threshold
was
50
dB
SPL
in
this
frequency
range.
In
different
animals
neurons
of
the
morphologically
uniform
type
AN1
were
found
to
have
marked
physiological
differen-
ces.
Narrow-band
neurons
with
threshold
curves
176
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
Intensity
[dB
SPL]
BNC
la
I
2
5
1
1
0
20
90-
70
50
30
-
90]
BNC
id
I
Is
I
lh
l I
70-
oh
I
5
10
20
Frequency
[kHz]
confined
between
4
and
8
kHz
were
found,
as
well
as
neurons
with
such
broad-band
responses
that
all
the
frequencies
tested
were
suprathreshold
at
high
intensities.
AN2
neurons
were
most
sensitive
to
14-16
kHz
the
carrier
frequency
of
the
courtship
song
at
35-40
dB
SPL.
Again,
though
morpholo-
gically
uniform,
AN2
neurons
in
different
individu-
als
exhibited
large
differences
in
their
threshold
curves.
Usually
the
curve
was
bimodal,
with
great-
est
sensitivity
at
15
kHz
and
a
secondary
maximum
at
5
kHz,
the
latter
being
more
or
less
prominent
in
different
animals.
Not
uncommonly,
however,
anatomically
identified
AN2
neurons
had
broad-
band
threshold
curves
with
the
thresholds
to
5
and
15
kHz
no
more
than
5
dB
below
those
to
the
rest
of
the
frequency
range.
The
BNC1
category
also
comprised
both
low-
frequency
and
high-frequency
cells.
The
low-fre-
Fig.
3.
Response
maps
of
two
BNC]
neu-
rons.
In
the
reconstructions
and
schemat-
ical
drawings
(inset)
on
the
right,
note
that
the
arborization
regions
overlap
the
pro-
jection
fields
of
ascending
neurons
(com-
pare
with
AN]
and
AN2
in
Fig.
2).
The
BNC]
class
includes
both
low-frequency
(upper
map)
and
high-frequency
to
broad-
band
neurons
(lower
map)
quency
BNC1
neurons
have
threshold
curves
simi-
lar
in
shape
to
those
of
the
AN1
neurons,
but
shifted
to
higher
intensities
throughout
the
fre-
quency
range
(Figs.
2,
3).
There
was
no
sign
of
sharpening
of
the
threshold
curve
of
BNC1
as
com-
pared
to
AN1.
Other
BNC1
neurons
had
threshold
curves
like
those
of
AN2,
with
maximal
sensitivity
at
15
kHz
and
a
second
maximum
at
5
kHz;
as
in
AN2,
the
5
kHz
component
varied
in
size
from
animal
to
animal.
The
high-frequency
BNC1
neu-
rons
also
had
higher
thresholds
at
all
frequencies
than
the
AN2
neurons.
The
BNC2
neurons
were
all
maximally
sensitive
at
5
kHz.
Two
types
could
be
distinguished.
One
of
them
was
extremely
sharply
tuned
to
5
kHz,
fre-
quencies
outside
the
range
4-6
kHz
eliciting
no
response
even
at
the
highest
intensities.
This
type
was
thus
more
sharply
tuned
than
either
AN1
or
50-
30-
177
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
Intensity
[dB
SPL]
90
-
BNC
2a
•I
70
-
50
-
30
-
psi
2
5
10
20
BNC
2c
90
-
70
-
50-
30-
2
5
10
20
Frequency
[kHz]
BNC1.
The
second
type
had
a
broader
threshold
curve
with
a
second
sensitivity
maximum
at
15-16
kHz
(Fig.
4).
Even
at
the
best
frequency
the
thresholds
of
the
BNC2
neurons
were
always
higher
(at
60-70
dB
SPL)
than
those
of
the
BNC1
and
AN
cells.
All
three
anatomical
classes
(AN,
BNC1
and
BNC2)
contained
neurons
having
their
lowest
threshold
at
the
carrier
frequency
of
the
conspecific
calling
song.
Moreover,
in
all
classes
neurons
were
found
with
two-peaked
threshold
curves,
tuned
to
both
the
carrier
frequency
of
the
calling
song
and
that
of
the
courtship
song.
Responses
in
the
suprathreshold
region
As
the
heights
of
the
bars
in
the
response
maps
(Figs.
2-4)
indicate,
at
suprathreshold
intensities
Fig.
4.
Response
maps
of
two
BNC2
neu
rons.
In
the
reconstructions
and
schemat-
ical
drawings
(inset)
on
the
right
note
that
the
arborizations
overlap
the
projection
fields
of
BNCI
neurons
but not
those
of
the
ascending
cells
(compare
with
Figs.
2
and
3).
BNC2
class
includes
both
sharply-
tuned
low-frequency
neurons
(upper
map)
and
neurons
with
a
broad
threshold
characteristic
(lower
map)
the
response
of
the
neurons
can
also
vary
with
frequency
and
intensity,
in
various
characteristic
ways.
To
assist
comparison
of
the
three
types
of
neurons,
response/intensity
curves
for
two
critical
frequencies
are
plotted
in
Fig.
5.
The
AN1
neurons
remained
mainly
tuned
to
the
carrier
frequency
of
the
conspecific
calling
song
even
at
suprathreshold
intensities
(Fig.
2).
The
in-
tensity
characteristic
of
AN1
neurons
rose
very
steeply,
following
almost
a
straight
line
between
50
and
70
dB
SPL.
At
intensities
above
70
dB
SPL
the
responses
of
AN1
neurons
with
broader
threshold
curves
continue
to
increase
sharply,
whereas
those
of
AN1
neurons
with
narrow
threshold
curves
de-
crease
(Fig.
5).
In
the
higher-frequency
range
(10-20
kHz)
the
spontaneous
activity
of
the
AN1
neurons
was
suppressed
at
intensities
about
20
dB
178
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
Spikes/Chirp
Spikes/Chirp
30
20
10
—•
AN
1
-•
AN
2
BNC
1
--.
BNC2
e
go
5
kHz
-
/
I/
30
-
20
I
/
10
A
—•
AN
1
--
-0
AN
2
o--O
BNC1
6
.
-
BNC2
15
kHz
fit
//
/
/
a
/
/
L
iT,
1
1
1
LAI
\
i
II
ce
7
/
/
/
Q
1
50
70
Intensity
[dB
SPL1
90
50
70
Intensity
[dB
SPL1
90
Fig.
5.
Intensity
characteristics
of
auditory
neurons
in
the
brain.
For
each
intensity
the
average
of
12
re-
sponses
is
shown
together
with
stan-
dard
deviation.
Left:
characteristics
of
low-frequency
neurons
and
AN2
stimulated
with
5
kHz.
Right:
charac-
teristics
of
high-frequency
neurons
and
AN1
stimulated
with
15
kHz
5
kHz
above
the
5-kHz
threshold.
This
inhibitory
side-
band
extended
from
50
to
75
dB
SPL.
Still
higher
intensities
produced
a
dramatic
change:
10—
to
20-kHz
stimuli
suddenly
elicited
marked
excitation.
This
stepwise
effect
of
increased
intensity
could
only
be
demonstrated
with
low-threshold
(35
dB
SPL)
AN1
neurons;
AN1
neurons
having
higher
thresholds
had
inhibitory
sidebands
that
extended
up
to
90
dB
SPL,
the
highest
intensity
tested.
All
AN2
neurons
had
intensity
characteristics
at
15
kHz
that
rose
steeply
without
saturating
over
the
entire
intensity
range
tested
(Fig.
5).
The
inten-
sity
characteristics
of
low-frequency
BNC1
neu-
rons
rose
considerably
less
than
those
of
the
corre-
sponding
ascending
neurons
and
always
eventually
saturated
(Fig.
5).
Two-peaked
BNC1
cells,
with
threshold
curves
like
those
of
the
two-peaked
AN2
neurons,
at
15
kHz
had
steeply
rising
intensity
cha-
racteristics
that
did
not
reach
saturation.
Low-frequency
BNC2
neurons,
like
the
corre-
sponding
BNC1
cells,
retained
their
tuning
to
4-6
kHz
in
the
suprathreshold
region
(Fig.
4);
however,
the
intensity
characteristics
of
the
BNC2
cells
were
considerably
flatter
than
those
of
the
other
auditory
interneurons.
At
intensities
more
than
10
dB
above
threshold,
the
responses
of
some
BNC2
neurons
no
longer
showed
any
dependence
on
intensity
(Fig.
5).
Latency
The
latency
was
defined
as
the
time
elapsed
bet-
ween
the
onset
of
the
stimulus
and
the
occurrence
40-
20
5
'
0
70
90
Intensity
[dB
SPL]
Fig.
6.
Latencies
of
individual
auditory
neurons
in
the
brain
for
5
kHz
stimulation.
For
each
intensity
the
average
of
12
respon-
ses
is
shown
together
with
standard
deviation
of
the
first
stimulus-correlated
action
potential.
The
range
of
latencies
was
16-38
ms
for
ascending
neurons,
22-42
ms
for
BNC1
neurons
and
38-79
for
BNC2
cells.
With
increasing
intensity
the
laten-
cies
of
all
neurons
became
shorter
(Fig.
6)
and
less
variable.
The
latency
shortening
was
most
pro-
nounced,
for
all
types
of
neurons,
at
about
ms
Latency
.—
AN
1
.---.AN
2
0-
-
0
BNC1
60
-
BNC
2
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
179
AN1
AN
2
BNC
1
c
BNC
2
c
Song
1
-
M-
80
dB
calling
I
IL_
Fig.
7.
Responses
of
various
auditory
brain
neurons
to
natural
songs.
Only
ANI
co-
pies
accurately
the
structure
of
the
chirps
in
the
calling
and
rivalry
songs.
The
response
of
ascending
neurons
to
the
first
syllable
in
a
chirp
outlasts
the
syllable
itself.
Calibration:
Abszissa:
20
ms,
Ordinate:
40
mV
(1st
and
2nd
trace),
20
mV
(3rd
and
4th
trace);
song
intensity:
80
dB
SPL
courtship
L
Chirp
3
9
AN
1
No.
9
-
3
-
BNC
la
9-
3
-
BNC
2o
90
180
90
18
'
0
90
180
n
8
6-
15
4-
3
5
1
1
90
180
90
180
90
180
Time
[ms1
Fig.
8.
Responses
of
various
auditory
brain
neurons
to
chirps
with
constant
temporal
structure.
Above:
each
action
potential
associated
with
the
consecutive
chirps
(bottom
to
top)
is
represented
by
a
dot.
Black
bars
represent
the
4
syllables
in
a
chirp
at
5
kHz
and
80
dB
SPL.
Below:
post-stimulus-time
histograms
of
the
same
responses
on
same
time
scale
(note
different
ordinate
scales)
10-20
dB
above
threshold,
and
above
80
dB
SPL
very
little
further
shortening
was
observed.
At
any
given
intensity
the
three
anatomical
types
differed
from
one
another
in
latency
by
more
than
5
ms,
which
allows
for
the
possibility
of
polysynaptic
connections
among
the
three
cell
types.
Responses
to
natural
songs
Typical
responses
to
the
three
main
kinds
of
songs
are
shown
in
Fig.
7.
AN1
copies
the
syllabic
struc-
ture
of
the
calling
and
rivalry
songs;
the
first
sylla-
ble
of
each
chirp
elicits
a
somewhat
longer
burst
of
impulses
than
the
others.
These
neurons
do
not
respond
to
the
courtship
song.
AN2
was
excited
by
all
three
songs.
There
was
a
detectable
grouping
of
the
discharge
in
the
pattern
of
the
calling
or
rivalry
song,
but
not
with
the
precision
found
for
AN1.
The
'tick'
sounds
in
the
courtship
song
excited
AN2,
which
responded
to
them
with
a
discharge
lasting
longer
than
the
syllables.
The
BNC1
neuron
in
Fig.
7
also
gave
a
clear
response
to
all
three
songs,
and
the
syllabic
structure
was
accurately
reflected
in
the
graded
potentials,
though
with
so-
me
syllables
the
latter
did
not
rise
high
enough
to
generate
an
action
potential.
Each
syllable
of
the
courtship
song
elicited
a
clear
response.
BNC2
neu-
rons
also
respond
to
the
calling
and
rivalry
songs,
though
the
degree
of
precision
in
copying
of
the
chirp
varied
considerably.
It
was
therefore
of
inte-
rest
to
analyze
their
responses
to
a
variety
of
beha-
viorally
effective
and,
in
particular,
behaviorally
ineffective
sound
patterns,
to
see
whether
their
tem-
poral
structure
had
a
specific
influence
on
the
pat-
tern
and
magnitude
of
the
responses
of
neurons
of
the
three
types.
Constancy
of
the
response
to
a
given
stimulus
pattern
When
synthetic
'chirps'
at
5
kHz
with
constant
temporal
parameters
known
to
elicit
positive
pho-
180
AN
1
BNC
1
a
BNC
1
d
BNC
2
a
'U
tLLtIIwL
j,
Sound
---0-0-0-0-1101:10-640-01.111---
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
IL
111.11L1L
I
L
Fig.
9.
Responses
of
various
auditory
brain
neurons
to
chirps
having
constant
energy
but
differing
temporal
structure.
Only
AN]
copies
all
the
patterns;
the
local
brain
neurons
have
specific
filter
proper-
ties.
Syllable
repetition
intervals
were
18
ms
(left),
34
ms
(middle)
and
98
ms
(right)
at
5
kHz
and
80
dB
SPL.
Calibra-
tion:
abszissa,
80
ms;
ordinate;
50
mV
(1st
and
2nd
trace),
30
mV
(3rd
and
4th
trace)
Time
Ims]
Synchronisation
BNC
2b
1-
BNC
2
120-
5
kHz
5
kHz
-
80
dB
80
dB
90
-
60-
30-
0-
iT
1-
BNC
1
5
kHz
80
dB
AN1
120
-
5
kHz
80dB
90-
60-
30-
0
1
1
I
;
26
50
74
98
26
Syllable
Repetition
Interval
Ems]
Fig.
10.
Copying
properties
of
various
auditory
brain
neurons
as
a
function
of
syllable
repetition
interval.
Left:
LCT
plots
(see
Materials
and
me-
thods)
for
single
neurons
in
each
of
the
three
classes;
after
latency
correction
the
time
of
occur-
rence
of
action
potentials
after
the
onset
of
each
syllable
was
plotted
on
the
ordinate
and
superim-
posed
for
each
syllable
during
the
chirp;
the
time
window
for
measurement
of
the
action
potentials
was
the
respective
syllable
period
plus
the
time
of
occurrence
of
the
following
action
potential
after
the
end
of
each
syllable
pause.
The
lower
dashed
line
shows
the
time
at
which
the
syllable
ends
for
each
SRI,
and
the
upper
dashed
line,
the
time
at
which
the
syllable
pause
ends.
Right:
synchroni-
zation
coefficient
as
a
function
of
SRI.
Each
con-
tinuous
line
connects
data
for
a
single
neuron;
se-
veral
neurons
of
each
class
are
represented.
Das-
hed
line
marks
5%
significance
level
(chi-square).
Sound
frequency:
5
kHz;
intensity:
80
dB
SPL
AN1
5
kHz
80dB
0-
'r
0
I I I
BNC
1
120-
5
kHz
80
dB
90-
60
-
30
-
0
Pr
11
.
1
.
11
50
74
98
notaxis
were
presented
repeatedly
at
constant
in-
tensity,
low-frequency
neurons
of
the
three
anato-
mical
types
responded
differently
(Fig.
8).
AN1
discharges
many
action
potentials
per
syllable,
and
the
responses
to
the
individual
syllables
are
dis-
tinctly
separated.
The
onset
of
the
response
to
each
syllable
in
the
chirp
occurs
at
a
constant
time
with
respect
to
that
syllable,
and
each
syllable
elicits
about
the
same
number
of
action
potentials;
there
is
no
habituation
of
the
response
during
repeated
stimulation
at
least
for
12
consecutive
chirps.
The
BNC1
neuron
in
Fig.
8
discharged
consi-
derably
fewer
action
potentials
per
chirp,
and
the
syllables
within
the
chirp
were
not
as
exactly
copied
as
by
AN1;
the
time
of
onset
of
the
responses
to
the
first
and
the
following
syllables
varied.
The
first
syllable
of
the
chirp
clearly
elicited
the
strongest
response,
and
there
was
a
sporadic
failure
of
the
response
to
one
of
the
following
syllables.
More-
over,
the
syllable
length
cannot
be
discerned
from
the
BNC1
response
pattern.
The
BNC2
neuron
in
Fig.
8
discharged
still
fewer
action
potentials
per
chirp.
There
is
a
response
to
the
onset
of
the
chirp,
though
its
latency
varies,
but
subsequent
syllables
frequently
fail
to
elicit
a
response;
indeed,
this
neuron
responded
to
no
more
than
three
of
the
four
syllables.
Though
there
might
be
some
degree
of
synchronization
of
the
response
with
the
syllable
pattern,
neurons
of
this
type
do
not
accurately
copy
the
syllabic
structure
or
chirp
length.
181
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
Responses
to
patterns
with
varied
temporal
structure
As
a
test
of
the
range
of
temporal
patterns
that
can
be
copied
by
neurons
of
the
various
anatomical
types,
constant-energy
chirps
(chirps
of
constant
length
and
repetition
rate
subdivided
into
syllables
of
various
durations,
always
with
50%
duty
cycle)
were
presented.
The
syllable
repetition
intervals
(SRI's)
within
these
chirps
varied
from
10
to
98
ms;
examples
are
shown
in
Fig.
9.
AN1
copied
the
syllables
over
a
wide
range
of
SRI's,
both
within
the
phonotactically
most
effec-
tive
region
and
on
either
side
of
it.
When
the
SRI
was
18
ms
or
more,
the
syllables
were
copied
very
accurately
by
80%
of
the
recorded
AN1
neurons.
Action
potentials
hardly
ever
occurred
in
the
pause
between
syllables,
and
the
onset
of
the
response
was
relatively
constant
with
respect
to
the
beginning
of
each
syllable.
These
features
of
the
response
are
more
clearly
evident
in
the
LCT
plots
(see
Me-
thods)
on
the
left
in
Fig.
10,
where
the
timing
of
the
action
potentials
associated
with
each
syllable
in
a
chirp
are
shown
superimposed
for
the
various
SRI's.
Grouping
of
the
BNC1
discharge
in
associa-
tion
with
individual
syllables
was
evident
only
with
very
long
SRI's,
although
some
temporal
pattern-
ing
of
the
response
could
be
discerned
even
with
shorter
SRI's.
Although
action
potentials
are
fre-
quently
discharged
during
the
pauses,
the
onset
of
the
BNC1
response
bears
a
relatively
constant
rela-
tion
to
each
syllable.
The
BNC2
neurons,
by
contrast,
show
respon-
ses
that
bear
practically
no
detectable
relation
to
the
individual
syllables.
In
general
there
are
more
action
potentials
during
the
syllables
than
during
the
pauses,
but
the
response
onset
varies
widely
with
respect
to
the
beginning
of
the
syllable
and
syllables
are
frequently
skipped
altogether.
To
this
imprecision
an
additional
effect
was
added
in
the
case
of
the
BNC2
neurons;
that
is,
more
action
potentials
are
discharged
in
response
to
chirps
with
particular
SRI's
than
to
others.
This
striking
pro-
perty
is
discussed
further
below.
As
a
numerical
expression
of
the
degree
to
which
the
chirp
structure
is
copied,
the
synchroni-
zation
coefficient
(SC)
offers
a
useful
measure
(see
Materials
and
methods).
SC
is
plotted
as
a
function
of
SRI
in
Fig.
10.
In
the
case
of
low-frequency
ascending
neurons,
synchronization
improves
as
SRI
increases.
Significant
synchronization
(P<
0.05)
was
found
for
these
cells
with
SRI's
of
18-25
ms
or
more.
The
SC's
of
BNC1
neurons
increased
far
less
rapidly
with
SRI,
reaching
the
5%
significance
level
either
not
at
all
or
only
with
SRI's
Spikes/Chirp
4-
BNC
2a
1
15
5
BNC
1a
26
50
74
98
40
-
20-
AN1
111111111111
26
50
74
98
Syllable
Repetition
Interval
[ms]
Fig.
11.
Magnitude
of
responses
of
various
auditory
brain
neu-
rons
to
chirps
varying
in
syllable
repetition
intervals;
note
diffe-
rence
in
ordinate
scale.
For
each
syllable
repetition
interval
the
average
of
12
responses
are
shown
together
with
standard
devia-
tions.
Sound
frequency:
5
kHz;
intensity:
60-80
dB
SPL
of
50
ms
or
more.
BNC2
neurons
exhibited
no
correlation
between
SC
and
SRI;
there
was
no
significant
synchronization
with
any
of
the
chirps
tested.
As
a
measure
of
the
intensity
of
the
response,
the
number
of
action
potentials
discharged
per
chirp
was
counted
for
each
of
the
various
constant-
energy
patterns
(Fig.
11).
For
a
given
pattern,
the
response
intensity
was
always
greatest
for
AN1,
followed
by
BNC1
and
BNC2.
With
stimuli
well
above
threshold
intensity,
the
strength
of
the
AN1
responses
did
not
depend
on
the
SRI.
At
lower
intensities
there
was
some
indication
that
SRI's
between
26
and
42
ms
elicited
stronger
responses
(t-
test,
0.05
<
P
<
0.1).
Thus
the
number
of
action
potentials
discharged
by
AN1
neurons
depends
chiefly
on
sound
intensity
and
not
on
SRI.
A
simi-
lar
result
was
obtained
with
many
BNC1
neurons;
that
is,
there
was
no
preferential
response
to
certain
SRI's
but
the
response
was
intensity-dependent.
By
contrast,
some
(see
Table
I)
BNC2
neurons
respon-
ded
more
strongly
to
chirps
with
SRI's
between
30
and
50
ms
than
to
those
with
longer
or
shorter
SRI's.
With
sound
intensities
greater
than
10
dB
above
threshold,
this
pattern-dependence
of
the
response
was
unaffected
by
intensity.
Moreover,
no
hysteresis
during
to-and-fro
scans
of
syllable
rate
dB
---•
80
---•
70
--•
60
25
Response
Magnitude
/
Chirp
%
75
Phonotax
is
O
`,
(
(:1
-]
c
o
l
o
\
/
i
/
11,
5
o
-75
/
0
0
0,
4
-25
182
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
1
1 1 1
ffi
i
ffi
26
50
74
98
Syllable
Repetition
Interval
[ms]
Fig.
12.
Relative
magnitude
of
responses
of
various
auditory
brain
neurons
to
chirps
varying
in
syllable
repetition
inter-
val.
The
stimulus
configurations
(con-
stant-energy
chirps
of
about
the
same
duration
but
with
different
SRI's)
are
illustrated
above
the
graph.
Each
of
the
three
curves
(maximum
response
=
100%
in
each
case)
shows
averaged
(12
stimulus
presentations)
responses
of
single
identified
neurons
with
particular
temporal
filter
properties.
Data
points
with
squares
were
from
neuron
BNC1d,
triangles
from
BNC2b
and
circles
from
BNC2a.
Open
symbols
(unconnected
by
lines)
show
responses
of
other
examples
of
these
indentified
neurons,
to
indicate
degree
of
variability.
Hatched
area
shows
relative
effectiveness
of
the
syllable
repetition
intervals
in
eliciting
phonotactic
tracking
(right
ordinate,
replotted
from
Thorson
et
al.
1982).
Sound
frequency:
5
kHz;
intensity:
80
dB
SPL
E
nor
habitutation
effects
(see
Materials
and
me-
thods)
were
visible
in
any
neuron
described
here.
Both
the
BNC1
and
BNC2
classes
also
contai-
ned
cells
that
responded
to
short
and
intermediate
SRI's
but
not
to
long
ones
(a
high-pass
filter
in
terms
of
syllable
repetition
rate),
and
others
re-
sponsive
to
long
and
intermediate
SRI's
but
not
to
short
ones
(a
low-pass
filter
in
terms
of
syllable
repetition
rate);
these
properties
are
illustrated
in
Fig.
12.
The
records
in
Fig.
9
suggest
that
the
low-
pass
characteristics
of
the
latter
neurons
result
from
temporal
summation
of
the
graded
poten-
tials;
the
relatively
long
syllables
at
intermediate
and
long
SRI's
permit
these
potentials
to
become
large
enough
to
generate
action
potentials,
whereas
with
shorter
syllables
they
do
not
built
up
to
the
threshold
level.
In
the
cricket
brain,
then,
there
are
neurons
with
responses
dependent
upon
the
temporal
struc-
ture
of
the
stimulus.
As
Fig.
12
shows,
neurons
with
low-pass,
high-pass
and
band-pass
characte-
ristics
were
detected.
The
pattern
specificity
of
the
latter
is
closely
correlated
with
that
for
patterns
that
elicit
the
animals'
phonotactic
behavior
(Thor-
son
et
al.
1982).
The
responses
of
these
neurons
were
maximal
for
syllable
periods
that
most
effecti-
vely
elicit
phonotaxis,
and
the
neurons
gave
weak
or
no
responses
to
phonotactically
ineffective
chirps.
Discussion
Anatomy
Two
ascending
cells,
here
stained
completely
for
the
first
time,
correspond
anatomically
to
the
neu-
rons
AN1
and
AN2
identified
in
the
prothoracic
ganglion
by
Wohlers
and
Huber
(1982)
and
apart
from
certain
differences
in
their
deutocerebral
ar-
borizations,
appear
identical
to
the
neurons
PALF1
and
PALF2
found
in
the
brain
by
Boyan
and
Williams
(1982)
but
not
to
the
HF1
neuron
described
by
Rheinlaender
et
al.
(1976).
All
the
ascending
neurons
so
far
identified
form
a
projec-
tion
field
in
a
narrowly
circumscribed
region
in
the
anterior
dorsal
protocerebrum,
lateral
to
the
a-lo-
be.
This
field
seems
to
be
specific
to
the
cricket,
for
ascending
auditory
neurons
of
Locusta
project
to
other
parts
of
the
brain,
chiefly
to
regions
in
the
lateral
protocerebrum
(Eichendorf
and
Kalmring
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
183
1980;
Boyan
1983).
The
main
projection
field
of
the
BNC1
neurons
overlaps
that
of
the
ascending
cells.
This
arborization
region
is
the
only
common
ana-
tomical
feature
of
the
BNC1
cells;
the
positions
of
their
cell
bodies
and
of
their
other
projection
re-
gions
vary
widely,
though
these
additional
arbori-
zations
are
usually
found
in
posterior,
lateral
or
median
regions
between
the
deuto-
and
protocere-
brum.
This
extensive
region
also
contains
arboriza-
tions
of
BNC2
neurons;
in
many
cases
these
neu-
rons
do
not
discernibly
project
into
two
distinct
regions.
No
comprehensive
description
of
these
parts
of
the
diffuse
neuropil
is
as
yet
available,
so
that
it
is
difficult
to
infer
what
anatomical
connec-
tions
the
BNC1
and.
BNC2
neurons
may
have
with
other
ascending,
local
or
descending
cells
having
projections
here.
Overlapping
has
been
observed
between
BNC1/2
neurons
and
local
multimodal
neurons
of
the
diffuse
neuropil
(Schildberger
1982),
as
well
as
descending
auditory
and
multimodal
cells
(Boyan
and
Williams
1981;
Schildberger
1982).
None
of
the
neurons
described
here
were
found
to
have
a
direct
connection
with
the
mushroom
or
central
body,
although
these
glomerular
neuropil
regions
receive
auditory
information
(Schildberger
1984).
Frequency
dependence
All
the
cricket
auditory
neurons
previously
descri-
bed
could
be
assigned
to
one
of
two
types
on
the
basis
of
their
threshold
curves:
low-frequency
neu-
rons,
sharply
tuned
to
the
carrier
frequency
of
the
conspecific
calling
song,
and
high-frequency
cells,
most
of
which
exhibit
broader-band
tuning
(Rhein-
laender
et
al.
1976;
Stout
and
Huber
1981;
Boyan
and
Williams
1982;
Popov
and
Markovich
1982;
Moiseff
and
Hoy
1983;
Boyd
et
al.
1984).
Boyd
and
his
coworkers
distinguished
two
different
low-fre-
quency
ascending
neurons
and
one
high-frequency
one.
In
the
present
study
only
one
high-frequency
and
one
low-frequency
type
were
found.
The
threshold
of
AN1
at
the
best
frequency
varied
be-
tween
35
and
50
db
SPL
in
different
animals.
In
some
cases
there
was
an
inhibitory
sideband,
and
in
others
not.
The
best
frequency
of
the
AN2
neu-
rons
was
15-16
kHz;
a
secondary
sensitivity
maxi-
mum,
at
5
kHz,
was
more
or
less
pronounced
in
the
various
neurons.
The
responses
of
AN2
neurons
strongly
varied
in
different
animals
in
respect
to
sound
frequency
and
pattern.
An
inhibition
of
high-frequency
ascending
neu-
rons
at
5
kHz
as
has
been
described
by
other
au-
thors
(Wohlers
and
Huber
1978;
Popov
and
Mar-
kovich
1982;
Boyd
et
al.
1984)
was
not
observed
in
this
study.
The
thresholds
of
the
brain
neurons
were
uni-
formly
higher
than
those
of
the
ascending
cells;
the
lowest
threshold
intensity
found
for
BNC2
neurons
was
55-60
dB
SPL.
In
behavioral
experiments
pho-
notaxis,
in
the
sense
of
maintained
tracking
of
the
sound
source,
is
not
elicited
by
intensities
below
ca.
50-55
dB
SPL
(Weber,
pers.
comm.).
But
the
abso-
lute
behavioral
threshold
(the
lowest
intensity
ca-
pable
of
eliciting
some
form
of
behavior)
is
un-
known,
so
it
remains
an
open
question
whether
the
thresholds
of
the
BNC2
neurons
are
equivalent
to
those
of
behavior.
Spatial
separation
of
the
high-
and
low-fre-
quency
processing
stations
in
the
brain,
as
discus-
sed
by
Boyan
(1980),
was
not
confirmed
in
these
experiments,
because
in
both
of
the
brain
regions
studied
here,
and
in
all
neuronal
classes,
low-fre-
quency
as
well
as
high-frequency
and
broad-band
neurons
were
found.
In
view
of
the
fact
that
ascending
low-frequency
neurons
can
be
inhibited
by
high-frequency
sounds
(Boyd
et
al.
1984)
and
high-frequency
neurons
by
low-frequency
sounds
(Boyan
1981)
in
two-tone
experiments,
one
might
infer
that
selective
switch-
ing
on
and
off
of
the
low-frequency
or
high-fre-
quency
pathway
occurs
in
various
behavioral
situa-
tions.
For
instance,
Boyan
(1981)
discusses
the
possibility
that
the
high-frequency
pathway
might
be
active
only
when
the
courtship
song
is
being
sung
and
suppressed
when
the
calling
song
is
pre-
sent.
However,
the
responses
of
the
high-frequency
neuron
AN2
to
the
calling
song
can
vary
from
strong
excitation
to
inhibition
in
different
animals
(Wohlers
and
Huber
1982).
Moreover,
the
high-
frequency
neuron
BNC
lc,
which
is
morphological-
ly
very
similar
to
the
neuron
PABN2
(Boyan
1980,
1981)
also
responded
to
the
calling
song.
So
there
might
be
morphologically
similar
but
physiologi-
cally
different
high-frequency,
but
also
low-fre-
quency
neurons
in
the
auditory
pathway
of
the
same
animal
or/and
physiological
variabilities
of
the
same
neuron
in
different
animals.
Intensity
The
characteristic
curves
of
the
ascending
low-fre-
quency
neurons
rise
sharply
as
straight
lines
over
a
range
of
about
30
dB,
whereas
the
intensity
cha-
racteristics
of
the
low-frequency
brain
neurons
stu-
died
so
far
rise
less
steeply,
and
those
of
most
BNC2
neurons
have
slopes
of
nearly
zero.
If
there
is
an
auditory
processing
mechanism
that
involves
these
neurons
in
the
order
AN-BNC1-BNC2,
one
184
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
could
therefore
expect
to
find
a
progressively
de-
creasing
intensity
dependence
of
the
response,
which
would
allow
to
discriminate
between
phono-
tactically
effective
and
ineffective
signals
indepen-
dent
of
intensity.
Temporal
pattern
Behavioral
experiments
on
phonotaxis
have
shown
that
sound
signals
elicit
a
positive
response
only
if
the
syllable
repetition
interval
of
the
chirp
is
in
the
range
of
30-60
ms
(Thorson
et
al.
1982)
a
range
that
includes
the
syllable
repetition
interval
of
the
natural
song.
Among
the
mechanisms
that
have
been
discussed
for
this
temporal-pattern
recogni-
tion
are
reciprocal
inhibition
(Wiese
1983)
and
a
Reiss
filter
(Thorson
et
al.
1982),
but
each
of
these
present
difficulties.
There
is
no
clear
evidence
that
the
temporal
dynamics
of
the
reciprocal
inhibition
of
the
omega
cells
in
the
prothoracic
ganglion
cau-
ses
the
cells
to
act
as
a
filter
that
passes
only
the
temporal
patterns
of
the
conspecific
signals
espe-
cially
in
view
of
the
fact
that
the
omega
cells
also
copy
phonotactically
ineffective
syllable
intervals
(Wohlers
and
Huber
1982).
A
recognizer
based
on
the
Reiss
filter
would
be
activated
not
only
by
the
conspecific
syllable
intervals
but
also
by
integral
multiples
thereof;
neither
in
the
behavior
(as
Thor-
son
et
al.
1982
point
out)
nor
in
the
neurons
studied
here
has
such
a
property
been
observed.
An
obvious
parameter
to
consider
in
the
search
for
neuronal
correlates
of
pattern
recognition
is
the
degree
of
synchronization
of
the
neuronal
dis-
charge
with
the
stimulus
pattern.
In
the
cricket,
AN
neurons
copy
the
temporal
structure
of
the
chirp
above
a
very
small
syllable
repetition
interval,
and
as
SRI
increases
the
response
becomes
progressive-
ly
more
synchronized.
However,
this
trend
conti-
nues
well
beyond
the
phonotactically
effective
ran-
ge;
that
is,
there
is
no
sign
of
optimization
of
synchronicity
in
the
natural
range
of
SRI's.
Brain
neurons
are
less
well
synchronized
with
the
stimu-
lus
pattern,
but
here
again
there
is
no
increase
in
synchronization
when
the
SRI
is
that
of
the
natural
song.
Indeed,
assuming
that
the
auditory
informa-
tion
passes
from
the
ascending
neurons
to
the
BNC1
to
the
BNC2
neurons,
there
is
a
progressive
decrease
in
the
accuracy
of
copying
along
this
path-
way.
There
is
no
evidence
here,
then,
for
recogni-
tion
based
on
synchronization
with
a
specific
sti-
mulus
pattern,
nor
has
such
a
recognition
mecha-
nism
been
found
in
other
studies
of
cricket
brain
neurons
(Boyan
1980).
However,
the
intensity
of
the
BNC2
response
the
number
of
action
potentials
discharged
per
chirp
does
depend
on
the
SRI,
and
is
often
maxi-
mal
for
intervals
that
are
phonotactically
effective.
That
is,
these
BNC2
neurons
are
band-pass
filters
of
syllable
rate
with
characteristics
very
close
to
the
band-pass
characteristics
of
the
behavioral
respon-
se.
Moreover,
brain
neurons
have
been
described
here
that
could
act
as
high-pass
and
low-pass
tem-
poral-pattern
filters.
A
BNC2-type
band-pass-filter
response
could
arise
via
the
ANDing
of
the
respon-
ses
of
these
two
kinds
of
neurons.
A
similar
situa-
tion
has
been
found
in
the
toad
brain,
where
there
are
auditory
neurons
in
which
no
specific
synchro-
nization
occurs,
but
the
discharge
rate
can
be
maxi-
mized
by
presenting
sounds
with
particular
tempo-
ral
structures;
these
include
neurons
with
high-,
low-
and
band-pass
characteristics
(Capranica
and
Rose
1983;
Rose
and
Capranica
1984).
It
appears,
then,
that
the
AN,
BNC1
and
BNC2
neurons
in
the
cricket
might
constitute
a
system
in
which
information
as
to
the
occurrence
of
a
conspe-
cific
calling
song
is
converted
in
steps
from
an
accurately
copied
representation
to
a
temporally
filtered,
intensity-independent
signal
capable
in
principle
of
triggering
the
appropriate
behavioral
response.
Acknowledgements.
I
thank
Drs.
M.
A.
Biedermann-Thorson,
F.
Huber,
J.
Thorson
and
T.
Weber
for
helpful
discussions
and
criticism
of
the
manuscript,
Dr.
Biederman-Thorson
for
transla-
tion
into
English
and
Dr.
H.
U.
Kleindienst
for
help
with
the
experimental
setup.
I
also
thank
Ms.
H.
Bamberg
and
Ms.
M.
L.
Obermayer
for
their
assistance
with
the
photography
and
Ms.
K.
Meyer
for
her
help
with
the
histology.
References
Boyan
GS
(1980)
Auditory
neurones
in
the
brain
of
the
cricket
Gryllus
bimaculatus
(De
Geer).
J
Comp
Physiol
140:81-93
Boyan
GS
(1981)
Tow-tone
suppression
of
an
identified
audito-
ry
neurone
in
the
brain
of
the
cricket
Gryllus
bimaculatus
(De
Geer).
J
Comp
Physiol
144:117-125)
Boyan
GS
(1983)
Postembryonic
development
in
the
auditory
system
of
the
locust.
J
Comp
Physiol
151:499-513
Boyan
GS,
Williams
JLD
(1981)
Descending
interneurones
in
the
brain
of
the
cricket.
Naturwissenschaften
67:486
Boyan
GS,
Williams
JLD
(1982)
Auditory
neurones
in
the
brain
of
the
cricket
Gryllus
bimaculatus
(De
Geer):
Ascending
in-
terneurones.
J
Insect
Physiol
28:493-501
Boyd
P,
Kiihne
R,
Silver
S,
Lewis
B
(1984)
Two-tone
suppres-
sion
and
song
coding
by
ascending
neurons
in
the
cricket
Gryllus
campestris
L.
J
Comp
Physiol
A
154:423-430
Capranica
RR,
Rose
G
(1983)
Frequency
and
temporal
proces-
sing
in
the
auditory
system
of
anurans.
In:
Huber
F,
Markl
H
(eds)
Neuroethology
and
behavioral
physiology.
Springer,
Berlin
Heidelberg
New
York
Tokyo,
pp
136-152
Casaday
GB,
Hoy
RR
(1977)
Auditory
interneurons
in
the
cricket
Teleogryllus
oceanicus:
Physiological
and
anatomical
properties.
J
Comp
Physiol
121:1-13
Eichendorf
A,
Kalmring
K
(1980)
Projections
of
the
auditory
ventral-cord
neurons
in
the
supraesophageal
ganglion
of
Locusta
migratoria.
Zoomorphologie
94:133-149
Esch
H,
Huber
F,
Wohlers
DW
(1980)
Primary
auditory
neu-
K.
Schildberger:
Auditory
neurons
in
the
cricket
brain
185
rons
in
crickets;
Physiology
and
central
projections.
J
Comp
Physiol
137:27-38
Hutchings
M,
Lewis
B
(1981)
Response
properties
of
primary
auditory
fibers
in
the
cricket
Teleogryllus
oceanicus
(Le
Guil-
lou).
J
Comp
Physiol
143:129-134
Kleindienst
HU,
Koch
UT,
Wohlers
DW
(1981)
Analysis
of
the
cricket
auditory
system
by
acoustic
stimulation
using
a
clo-
sed
sound
field.
J
Comp
Physiol
141:283-296
Moiseff
A,
Hoy
R
(1983)
Sensitivity
to
ultrasound
in
an
identi-
fied
auditory
interneuron
in
the
cricket:
Possible
neural
link
to
phonotactic
behaviour.
J
Comp
Physiol
152:155-167
Nocke
H
(1972)
Physiological
aspects
of
sound
communication
in
crickets
(Gryllus
campestris
L.).
J
Comp
Physiol
80
:
141-162
Pollack
GS,
Hoy
R
(1979)
Temporal
pattern
as
a
cue
for
species-
specific
calling
song
recognition
in
crickets.
Science
204:
429-432
Pollack
GS,
Huber
F,
Weber
T
(1984)
Frequency
and
temporal
pattern
dependent
phonotaxis
of
crickets
(Teleogryllus
ocea-
nicus)
during
tethered
flight
and
compensated
walking.
J
Comp
Physiol
154
A:13-26
Popov
AV,
Markovich
AM
(1982)
Auditory
interneurons
in
the
prothoracic
ganglion
of
the
cricket,
Gryllus
bimaculatus.
J
Comp
Physiol
146:351-359
Popov
AV,
Shuvalov
VF
(1977)
Phonotactic
behaviour
in
crick-
ets.
J
Comp
Physiol
119:111-126
Popov
AV,
Markovich
AM,
Andjan
AS
(1978)
Auditory
inter-
neurons
in
the
prothoracic
ganglion
of
the
cricket,
Gryllus
bimaculatus
DeGeer.
J
Comp
Physiol
126:183-192
Rheinlaender
J,
Kalmring
K,
Popov
AV,
Rehbein
H
(1976)
Brain
projections
and
information
processing
of
biologically
significant
sounds
by
two
large
ventral-cord
neurons
of
Gryl-
lus
bimaculatus
De
Geer
(Orthoptera,
Gryllidae).
J
Comp
Physiol
110:251-269
Rose
G,
Capranica
RR
(1984)
Processing
amplitude
modulated
sounds
by
the
auditory
midbrain
of
two
species
of
toads:
matched
temporal
filters.
J
Comp
Physiol
154
A:211-
219
Schildberger
K
(1982)
Untersuchung
zur
Struktur
and
Funk-
tion
von
Interneuronen
im
Pilzkorperbereich
des
Gehirns
der
Hausgrille
Acheta
domesticus.
Dissertation,
Gottingen
Schildberger
K
(1984)
Multimodal
interneurons
in
the
cricket
brain:
Properties
of
identified
extrinsic
mushroom
body
cells.
J
Comp
Physiol
154
A
:71-79
Stout
JF,
Huber
F
(1981)
Response
to
features
of
the
calling
song
by
ascending
auditory
interneurons
in
the
cricket
Gryl-
lus
campestris.
Physiol
Entomol
6:199-212
Thorson
J,
Weber
T,
Huber
F
(1982)
Auditory
behavior
of
the
cricket.
II.
Simplicity
of
calling-song
recognition
in
Gryllus,
and
anomalous
phonotaxis
at
abnormal
carrier
frequencies.
J
Comp
Physiol
146:361-378
Weber
T,
Thorson
J,
Huber
F
(1981)
Auditory
behavior
of
the
cricket.
I.
Dynamics
of
compensated
walking
and
discrimi-
nation
paradigms
on
the
Kramer
treadmill.
J
Comp
Physiol
141
:
215-232
Wiese
K
(1983)
Akustische
Kommunikation
bei
der
Grille:
Eigenschaften
der
bilateral
inhibitorischen
Kommissur
in
der
Horbahn.
Verh
Dt
Zool
Ges
1983:180
Wohlers
DW,
Huber
F
(1978)
Intracellular
recording
and
stain-
ing
of
cricket
auditory
interneurons
(Gryllus
campestris
L.,
Gryllus
bimaculatus
De
Geer).
J
Comp
Physiol
127:11-28
Wohlers
DW,
Huber
F
(1982)
Processing
of
sound
signals
by
six
types
of
neurons
in
the
prothoracic
ganglion
of
the
cricket,
Gryllus
campestris
L.
J
Comp
Physiol
146:161-173