Approaches to the cortical analysis of auditory objects
Approaches to the cortical analysis of auditory objects
复制标题
DOI：
10.1016/j.heares.2007.01.010
复制
发表时间：
2007-07-01
期刊：
影响因子：
2.8
通讯作者：
Friston, Karl J.
中科院分区：
文献类型：
--
作者：
Griffiths, Timothy D.;Kumar, Sukhbinder;Friston, Karl J.
我们描述的工作，解决了使用“通用”的声音，不对应于任何特定的事件或来源（如元音或语音），具有语义关联的听觉对象的分析的皮层基础。实验涉及操纵合成声音产生刺激特征的系统变化，如频谱包络.常规分析正常功能成像数据表明，频谱包络和感知的音色变化的分析涉及一个网络组成的颞平面（PT）的双边和右侧上级颞沟（STS）。使用动态因果模型（DCM）和贝叶斯模型选择的成像数据的进一步分析进行了右半球地区，以确定这些听觉区域之间的有效连接。具体地说，目标是确定网络中的频谱包络分析是以串行方式（即从HG到PT到STS）还是并行方式（即PT和STS同时接收来自HG的输入）完成的。选择了代表关于HG、PT和STS之间的连接性的不同假设的两个系列的模型，即串行和并行模型（总共16个）。在一个家庭内的模型不同的是由频谱包络的分析调制的路径。在模型被识别之后，贝叶斯模型选择过程然后被用于从指定的模型中选择“最优”模型。这些数据强烈支持一个特定的串行模型，其中包含调制的HG PT的有效连接在频谱包络variation.Parallel神经学科的工作，解决了病变的影响，这个网络的不同部分。我们最近详细研究了“听觉障碍”的主题：听觉对象感知质量的改变，与音高或响度变化不同。受试者具有上述正常网络的损伤，具有对音高强度的正常感知，但对频谱包络变化的分析的异常感知。(c)2007 Elsevier B.V.保留所有权利。
We describe work that addresses the cortical basis for the analysis of auditory objects using 'generic' sounds that do not correspond to any particular events or sources (like vowels or voices) that have semantic association. The experiments involve the manipulation of synthetic sounds to produce systematic changes of stimulus features, such as spectral envelope.Conventional analyses of normal functional imaging data demonstrate that the analysis of spectral envelope and perceived timbral change involves a network consisting of planum temporale (PT) bilaterally and the right superior temporal sulcus (STS). Further analysis of imaging data using dynamic causal modelling (DCM) and Bayesian model selection was carried out in the right hemisphere areas to determine the effective connectivity between these auditory areas. Specifically, the objective was to determine if the analysis of spectral envelope in the network is done in a serial fashion (that is from HG to PT to STS) or parallel fashion (that is PT and STS receives input from HG simultaneously). Two families of models, serial and parallel (16 in total) that represent different hypotheses about the connectivity between HG, PT and STS were selected. The models within a family differ with respect to the pathway that is modulated by the analysis of spectral envelope. After the models are identified, Bayesian model selection procedure is then used to select the 'optimal' model from the specified models. The data strongly support a particular serial model containing modulation of the HG to PT effective connectivity during spectral envelope variation.Parallel work in neurological subjects addresses the effect of lesions to different parts of this network. We have recently studied in detail subjects with 'dystimbria': an alteration in the perceived quality of auditory objects distinct from pitch or loudness change. The subjects have lesions of the normal network described above with normal perception of pitch strength but abnormal perception of the analysis of spectral envelope change. (c) 2007 Elsevier B.V. All rights reserved.