77.1
77.2
IIGVEIISlZ NEITROETIIOI.O(iY: I IO\\' 1)OES BBIIAVlOR INI-ZLIENCBTHE
NB1IVC)IIS SYSTEM? IT. A. I ~ ~ I ~ I I Sti~niwJ
I ~ I I I I U~~ivcrsity,
~~,
Dep:utmcnl of
hyclii~logy,Sordiu~Ilidl, Bldg. 420, S~:u~lwd.
CA 04305.
R o b u s t n e s s of a m p l i t u d e m o d u l a t i o n e n c o d i n g b y P - r e c e p t o r
afferent s p i k e t r a i n s of weakly electric fish.
Gabriel ~ < r e i ~ n a n ' * , ~ a b r iGahbiani',Krallrr
zio
~ I ~ t ~ n e ~ ~ . ~ Ko(.lii
lu~istof
I Computation and Wrural Syst,ems.Caltrcli.Pasi~r1~~1ia.C~4
'31125
Dept, of Biology,UCR.Riverside.CA 92521
P-type tuberons reccpt,ors in weakly eI<wtric gyninotiihrnu fish i ~ i code rllanges in electric fidd arnplitrrdr~.Singlr rrrrptor affi~icritsLriiirsnit high infor~natiol~
rates ahont random amplitndc ruoduialvil slii~uiili
(T;Vrssel r t al. 1996, J. Nezirophys. 75:2280-2293). We cllarnrtcrizivl
the rohustncss of enroding and its relationsl~ipt,o tllr varial,ility of 1'receptor afferents spike trains. Random a~nplitvdrrnodnlat,ed clrrtiir
fields wrre presentrd to curarized fish wluilr recording extrarcllnlmly
from P-rc~eptorafferents. Reconstruction trrhnicpies wrre ~ ~ s rtu( l rstimate the stimulus from the spikes. By adding random temporal jittw to
the spikes, we explored the timing precision rrquired for accurate cucocling. The robustness of encoding depended on stinullus cut-ofi frrqurwy
(f,) and contrast. The amount of jitter requirrd for a 50 O/o drop in
coding fraction was 3 111s for high f, (88Hz) but it was higher tluau 'LO
111s for low f, (less than 10 Hz). Similar results were obtainrd by random deletions and additions of spikes. By reprating the sarnr 'fromi!'
st,inuuios 10 times, wc a~ualyzcdthe Iinie windows ovrr wltich tl~r.s11ikr.s
arc reproducible using spikr xnetrics (Virtor rt al. 1996, .I.Neu~ophys.
76:1310-1326). Vmiability decreased for high cnt-ofl frcqurncic~shnt n-c
did not observe spike alignment a t i.11~ins level in rast,cr plots Ir! spitr
of this, the intertrial variance versus meall spike count rlnstrrrd near tlic
curves of minimum variance. Our results snggrst that P-recept,or afkrrnts nse an adaptive coding scl~e~ne.
Accurate aud mbnst inforniation
can be carried for low f, withont spike time relial~ilitya t the IIIS lrvcl
while at high f, the code is not lobust to jittrriug at t,be 3 ins Irvrl.
Supported hy NSF and ERC
'
77.3
77.4
BLOCKING GAP JUNCTIONS IN THE ELECTRIC FISH PACEMAKER NUCLEUS ALTERS THE FREQUENCYAND SYNCHRONY OF OSCILLATIONS.
K. T. Moortgatl.?*, T. H. Bullock2, and T. J. Sejnow~ki'.~. 'Computational
Neurobiology Laboratory, The Salk Institute and ?University of California at
San Diego, La Jolla, CA 92037.
The medullary pacemaker nucleus (Pn), which commands the electric organ discharge of a wave-type gymnotiform electric fish, is the most regular biological oscillator known. We have investigated the relative importance of its
networkand cellular mechanisms in setting its temporal regularity. The ca. 150
Pn neurons, which fire synchronously and command the EOD on a cpcle-bycycle basis, are thought to be coupled only electrotonically, with no chemical
synapses intrinsic to the Pn. To investigate the mechanism of the electromotor oscillations and their extreme regularity, we blocked gap junctions in an
in uitro preparation of the Pn. Normally, the neurons of the PII are both frequency and phase-locked, and fire every cycle of the constant-frequency oscillation though with phase differencesof up to ca. 250 ps as measured with dual
intracellular recordings. After application of halothane (2 - 5%),a gap jmction blocker, the frequency of intracellular and local field potential oscillations
decreased steadily to ca. 50% of their original. At the same time, the cells remained synchronized, possibly due to incomplete block. However, the relative
phase between cell pairs or between a cell and a field potential shifted, even to
the point of changing sign. Further halothane application silenced the Pn neurons' spikes. Both the long and short term effects are reversible. Thus we find
that regularity and frequency locking to some degree survive partial block and
cannot be entirely dependent on network integrity. We further compare the effects of halothane to those of other gap junction blockers. Support: NIHM fellowship to KTM; and HHMI to TJS.
PARTICIPAllCiiCF THETORUSSEMtCRCULARlSWRSALlS IN THE MODULATION
OF BEHAVIORS THROUGH ACIWG OF GABAERGIC S Y m IN AN EKCIRIC
FISH. G m o h o cmnpo T T Dtmi@ S A L C o k A S F % m a and A H o f f iLa,o;
muw@ah excitability. 2) W l i n e EOD and the excitability of the animals did noi &Em
teihre and after administmtionof GABA aeonistic muscimol and saline Chus d t s suprts;
hat TSd mntmls EGDarresfs ~ ~ c hagomc e n a ~ u l t ethrought
~
the t a n ~ d &
y
GABA~csystnn
Suplnmiky FAPEsP(97/01815-3)
i
77.5
77.6
Rate codina of sound features in second and third order auditorv
nuclei is computed from temporally structured first order spike trains.
AUDITORY TEMPORAL COMPUTATION: A SIMPLE
NETWORK MODEL FOR FEATURE EXTRACTION FROM
COMMUNICATION SOUNDS. R.H. Cudtnore'. E.W. Lareez. J.D.
Crawford2'. 'Dept. of Cotnputcr and Information Science; 'Dept. of
Psychology, University of Pennsylvania, Philadelphia, PA 19104.
Temporal features of sounds carry information In human speech
and in the coininunication systems of a variety of animal species.
Temporal measurements on the scale of one to hundreds of rns arc required
of the auditory nervous system for the analysis of con~municationsounds.
A number of authors h a w postdated a role for inhibition, and i~rh~bitory
rebound excitation, in c~rcuits that function as temporal filters for
extracting salient feahlres from sounds (c.g. Margoliash 1983; Olsen and
Suga 1991; Casseday et a1 1994). We have been investigating the role of
inhibitory gating inechanisn~s in auditory midbrain circuits that yield
sensitivity to temporal features of comn~unicationsounds used by the fish
Pollimynrs. We have developed a dynamic computational model that
takes spike trains highly synchronized to an auditory stimulus as mpnt, and
produces temporally tuned output based on the temporal dynamics of
auditory midbrain.
T l ~ ekinetics of both the inhibitory post
neuron. WIH R 0 1 DC01252-041.