Characteristics And Mechnisms Of The Processing Of Complex Time-frequency Information In Sub-cortical Auditory Pathway

Natual sounds usually contain complicated context, sources and temporalproperties, which demands the auditory system of animals and humans to encode andprocess the complex spectral components and fast time-varying signals in a rapid andprecise way. The research for the processing of complex auditory information is stillin an early stage. This study focuses on the characteristic and mechanisms of coding,adaptation and dynamics of complex sound of neurons in the rat inferior colliculus(IC), using complex time seqeunce, complex spectrum and time-frequency combinedmethods.First, oddball paradigm is used to investigate the novelty detection andstimulus-specific adaptation (SSA) in rat IC. Both experiments and data analyses aredone systematically. From data of over80neurons, it is confirmed that firing rate andfirst-spike latency (FSL) both demonstrate SSA effect for IC neurons. Moreover, thedynamics of responses during the oddball sequence is analyzed. Long-term adapationis found and fitted by exponential and power functions. The difference of long-termadaptation of firing rate and FSL is compared by their time constants. The short-termdynamics is also analyzed. The contribution of local stimulus history and globalprobability to SSA is investigated by a linear model. In addition, responses to soundlevel and amplitude modulation difference is examine and no significant SSA isfound.Second, the neural resposne to sounds with complicated spectrum structure istested with Random Spectra Stimuli (RSS). The first-order RSS model and theirweights are utilized to describe the frequency selectivity of neurons. It is found thatfor most IC neurons RSS is suitable for describing the spectral receptive field of theneuron. Two prediction factors are calculated and used to quantitatively evaluate theRSS approach.Third, for the dependency of spectro-temporal receptive field (STRF) ondifferent sounds that many experiments have observed, a linear system model is buildup in this study to explain the shape of STRF and their variations with sound intensity. Under the optimization of mutual information and neural cost, the optimal filterparameters under different sound statistics are derived. The model predicted STRF iscompared with experiment result and our simulation shows that the change of STRFwhen sound intensity increases qualitatively matches observations from experiment.In addition, a lot of improvements are made for the experiment platform in thisstudy. The technique of lesion and stain of brain slices are modified and have becomemore stable and feasible. More importantly, a new platform for recordings on awakehead-fixed rats are being developed and improved.