La lecture à portée de main
Découvre YouScribe en t'inscrivant gratuitement
Je m'inscrisDécouvre YouScribe en t'inscrivant gratuitement
Je m'inscrisDescription
Sujets
Informations
Publié par | technische_universitat_munchen |
Publié le | 01 janvier 2010 |
Nombre de lectures | 28 |
Langue | English |
Poids de l'ouvrage | 77 Mo |
Extrait
Auditory
Information
cessingPro
withSystemsin
ternallyIn
Coupled
Christine
oßenV
Ears
This
dissertation
has
memoirtheingclass.
pdfLATEX.
eenb
written
in
aswesettingypT
us-
LATEXus-
withdone
FundinghasbeenprovidedbytheBernsteinCen-
terforComputationalNeuroscience(BCCN)–Mu-
h.nic
TECHNISCHELehrstuhlf¨urUNIVERSITTheoretisc¨heATPhM¨ysikUNCHEN
ProInformationAuditorycessingwithSystemsinEarsCoupledternallyIn
oßenVChristine
Vollst¨andigerAbdruckdervonderFakult¨atf¨urPhysikderTechnischenUniversit¨at
M¨unchenzurErlangungdesakademischenGradeseinesDoktorsderNaturwissenschaften
Dissertation.genehmigten
Vorsitzender:Univ.-Prof.Dr.F.Simmel
Pr¨uferderDissertation:1.Univ.-Prof.Dr.J.L.vanHemmen
BornemannF.Dr.Univ.-Prof.2.
DieDissertationwurdeam12.07.2010beiderTechnischenUniversit¨atM¨uncheneingereicht
unddurchdieFakult¨atf¨urPhysikam27.07.2010angenommen.
.
.
.
es
sind
andegenst¨Ge
uns
Dinge
Sinneernseru
als
außer
alen,gebge
uns
efindb
liche
wasdem,vonlein
sieansichselbstseinm¨ogen,wissenwirnichts,sondern
kennen
die
sie
nurihreErscheinungen,d.i.dieVorstellungen,
in
uns
wirken,
indemsieunsereSinneaffizieren.
uelImman
1804)–(1724tKan
Prolegomena,
1783
tenConts
viitstenCon1ductiontroIn11.1Soundstimuli...................................2
1.2Mechanicalprocessingofsoundstimuli.....................3
1.2.1Evolutionofdifferentauditorysystems.................3
1.2.2Internallycoupledears..........................3
1.3Neuronalprocessingofsoundstimuli......................5
1.3.1Buildingblocksofneuronalcomputation................5
1.3.2Leakyintegrate-and-fireneuron.....................9
1.3.3Poissonneuron..............................10
1.4Neuronalrepresentationofsoundstimuli....................13
1.4.1Neuronalmaps..............................13
1.4.2Pre-wiringamap.............................13
1.4.3Fine-tuningamap............................14
2ModelingInternallyCoupledEars:TheICEmodel17
2.1Introduction....................................17
2.2DerivationoftheICEmodelwithcylindricalmouthcavity.........21
2.2.1Externalsoundinput...........................22
2.2.2Internalcavity..............................23
2.2.3Vibrationofthemembrane.......................25
2.3Numericalsimulationoftheeigenfunctions
ofrealisticmouthcavities............................31
2.4Evaluationandresults..............................33
2.4.1Directionalityofthemembranevibrationpattern...........34
2.4.2Eigenmodesofarealisticmouthcavity.................39
2.4.3GeneralizationoftheICEmodel....................42
2.4.4Spatialvibrationpatternofthemembrane...............49
2.5Conclusion....................................50
3NeuronalprocessingofiTDsandiADs53
3.1Introduction....................................53
3.2SeparatedpathwaysforiTDandiADprocessing...............54
3.3ProcessingofiTDs................................55
vii
4
5
6
tstenCon
3.4ProcessingofiADs................................62
3.4.1Experimentalandnaturalstimulus...................63
3.4.2Simulationofnucleusangularis(NA)..................64
3.4.3SimulationofEIneurons........................65
3.5Conclusion....................................69
AuditorySensitivityandInternallyCoupledEars73
4.1Introduction....................................73
4.2Theoreticaldescriptionofmembranevibrations................75
4.2.1Membranevibrationdifferencesofindependentears.........77
4.2.2Membranevibrationdifferencesofinternallycoupledears......78
4.2.3Empiricalanalysis............................79
4.3Conclusion....................................82
4.AMethods:Empiricalanalysis...........................82
Optimalityinmono-andmultisensorymapformation87
5.1Introduction....................................87
5.2Fundamentalconceptofneuronalmaps....................89
5.3Mathematicalmodel...............................91
5.3.1Definitionoftheproblem........................92
5.3.2Optimalreconstruction.........................94
5.3.3Matrixnotation.............................97
5.3.4Relationtothemaximum-likelihoodapproach.............98
5.3.5Neuronalrealizationofthemodel....................99
5.4Arecipeofmakingmaps.............................100
5.5Multimodality...................................101
5.5.1Multimodalinteraction.........................101
5.5.2Developmentofmultisensoryspace...................102
5.6Discussion.....................................104
5.ANonlinearitiesininformationprocessing....................105
5.BSelf-averaging...................................105
5.CRemainingderivationstepsleadingto(5.23)..................106
5.DGaussianblurredsignal.............................109
111formationmapdalMultimo6.1Introduction....................................111
6.2Theintegratedmultimodalteacher.......................114
6.2.1HowdounisensorymapsdetermineiMT?...............114
6.2.2HowdoiMTcharacteristicsinfluencemapadaptation?........118
6.2.3HowdoesiMTcalibratedifferentunimodalmaps?..........122
6.3ApplicationsoftheiMTconcept........................122
6.3.1Experimentspro-vision-guidedmapformation.............122
6.3.2Experimentscontra-vision-guidedmapformation...........126
6.4Discussion.....................................128
6.AOptimalcombinationoftwomodalities.....................129
viii
7
tstenCon
6.BOptimalcombinationofthreemodalities...
6.CModelParameters...............
tlyrequenF
yBibliograph
used
abbreviations
and
functions
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
130131
133
135
ix
ductiontroIn1.
Ajourneyofathousandmilesmustbegin
step.esinglawithLaoTzu
Thischapterintroducesthefundamentalsofsensoryinformationprocessing,focusingon
theprocessingofsoundsignalsastheauditorysystemisoneofthemostwidespreadof
thevarioussensorysystems.Theprominentroleoftheauditorysystemcanbeexplained
bythespecificpropertiesofsoundperception,namely,hearingisomnidirectionalandits
processingisveryfast.Incomparisontolight,soundwaveshaveamuchlongerwavelength
sotheyarenotblockedbysmallobjects.Wecan,forinstance,hearsomethingbehinda
treebutwecannotseeit.Amongstotheradvantagesauditionthereforeoffersthepossibility
toreacttoapproachingdangersthatarenotyetvisible.Toexploitthelatteradvantage,
itisessentialthatananimalcanlocalizeasoundsource,otherwisethepreycouldtryto
escapeinthedirectionofthepredator.Todeterminethedirectionofasoundstimulus,
severalstepsarenecessaryintheanimalsstudiedinthisthesis.
First,anobjectevokesanauditorystimulus.Section1.1describeshowsuchastimulus
propagatesthroughthesurroundingmediumandisamplifiedbeforearrivingatthe
animal.theofdetectorsSecond,thearrivingsoundwavesexcitethetympani,apairofthinmembranesthat
arepartofthemechanicalauditorysystem.Theanatomyoftheauditorysystemishighly
variable.Theprimaryfocusofthisworkisoninternallycoupledears(ICE)inwhichone
membranecaninfluencevibrationoftheotherthroughinternalcavities;seeSec.1.2.
Third,tympanicvibrationsareprocessedneuronally.Thefundamentalelementof
neuronalcomputationistheneuronalnet,consistingofneuronsasbuildingblocksand
thevariableconnectionsbetweenthem,calledsynapses.Section1.3reviewsthegeneral
functionandmathematicalmodelingofneuronsandsynapses.
Fourth,asresultofneuronalcomputation,sensorystimuligiverisetoneuronalrep-
resentationsofthestimuli,i.e.,neuronalmaps;seeSec.1.4.Eachneuronofthemap
representsaspecificproperty,e.g.,thestimulusataspecificpointinspace.Neighboring
neuronsrespondtosimilarsensoryinputs.Neuronalmapsreconstructthestimulusaswell
1
troIn1.duction
aspossible,i.e.,optimallywithinthelimitsofprocessing.Theprecisecalibrationofthe
synapsesrequiredforstimulusreconstructionresultsfromanexperience-basedlearning
processthattakesintoaccountinputsfromallavailablesensorysystems.
ulistimSound1.1Soundwavesareoscillationsofpressurethataretransmittedthroughtheair.Asound
stimulusiscreatedwhenamovementcompressesthesurroundingairandcreatestraveling
soundwavesthatarriveatthetympaniofananimal.Dependingonthespatialrelationship
betweenthetwotympaniandthesoundsource,thearrivingsoundwavesdifferinphaseand
amplitudebetweenthetympani.Thesedifferencesaredenotedasinterauraltimedifferences
(sideITDofs)theandinhead,terauralforinstance,amplitudearrivesatdifferencesthe(facing,IADs).i.e.A,soundipsilaterwalavtefromympanaumsourceearlieratandone
withhigheramplitudethanattheaverted,i.e.,contralateraltympanum.Asoundwave
fromasourceinfrontofthehead,incontrast,arrivesatbothearssimultaneouslyandwith
thesameamplitude.ThesetwoexamplesdemonstratethatITDsandIADscanencode
thedirectionofasoundsource.AnevaluationofITDsandIADsasperformedbythe
auditorysystemandtheconsequentneuronalcircuitscouldthereforedecodethesound
direction.source
2
Figure1.1:Evolutionofvertebrateears.Duringthetransitionfromwatertoland,tympanic
middleearscapableofreceivingairbornesoundevolvedseparatelyamongtheancestorsof
moderncross-sectionsfrogs,throughturtles,diffeli