Bistability and alternating streams during perceptual ambiguity

描述(由申请人提供)：作为一名公认的计算神经学家，我的目标是通过获得新的研究技能和知识来充实我的研究计划，使我能够阐述和回答与听觉感知相关的问题，特别是听觉流动及其神经生物学机制。自1973年获得博士学位以来，我的职业生涯涉及非线性动态系统概念的计算建模、仿真和应用，主要是在单元和电路级别，以确定单元和电路动力学的基本机制：集成和输入/输出属性，如节律性和双稳态。自20世纪90年代末以来，我一直将我的大部分资金投入到理解听觉时间处理的问题上--稳定地进行计算建模和(指导但不执行)体外膜片钳实验，主要是关于听觉脑干神经元的生物物理机制；它们显示出精湛的亚毫秒符合检测能力，用于神经元计算声音定位。最近，通过一次合作研究经验，我研究了模糊视觉场景的感知双稳模型。它是令人满意和成功的，它激励我获得知识和技能来研究听觉通道中的知觉场景问题。此外，虽然用于研究视觉感知双稳态的刺激通常是静态的，但在听觉通道中，刺激自然是动态的：声音在变化，因此“听觉对象”的结构特征或线索正在演变。我们如何及时识别和跟踪这些特征引发了关于动力学机制的引人注目的问题。我的动机是将动态系统概念和机械建模应用于流问题，我对感知/行为层面的研究充满热情，这与我过去的研究完全不同。从长远来看，我计划寻求资金，作为私人投资，用于招募、培训、与学生/博士后在听觉流媒体和知觉方面合作，以及建模和心理物理实验。我希望能够进入纽约大学现有的隔音音响室进行行为实验。具体地说，我计划在备受尊敬的听觉系统神经学家的指导下学习和工作：(1)获得基础知识、文学熟悉度、工作沟通技能，以及(2)获得如何设计、实施和监督听觉串流领域的心理物理实验和计算建模的实验知识。我的导师将是：阿尔伯特·爱因斯坦医学院的Elyse Sussman博士和马里兰大学Park学院的Shihab Shamma博士。我将主要从Sussman和她的工作人员那里了解用于流媒体的人类心理物理实验的设计、实现和数据分析，这样我就可以独立执行这些实验-包括我研究项目的实验部分。此外，我还将从Shamma那里了解到关于雪貂(听觉皮质)的神经生理学实验，以及他正在进行的流媒体研究的建模和数据分析。Shamma的专业知识对我的研究项目的建模部分将是有价值的，我的研究项目是开发流中双稳的机械模型。向两位导师学习背景和经验基础的过程将包括我阅读基础文献，定期与导师讨论，参加实验室会议，以及旁听实验。我计划参加每年由莫尼塔·查特吉教授的MD大学心理声学(实验听力学)研究生水平的课程。我的研究策略涉及两个相互关联的部分，计算模型和实验心理物理学，关于潜在模糊刺激的听觉流动动力学。刺激由两个不同频率(频差，df)的音调序列组成，并与呈现速率(PR)交织：A_B_A_A_……根据DF和PR的值，对象可以将A序列和B序列感知为不同的(分离的，两个流)或作为一个单元(集成的，一个流)；如果DF或PR较大，则为2个流。对于长时间的陈述，知觉可能会在时间上交替(持续几秒钟)(Pressnitzer&Hupe，2006)。我的主要假设是，双稳态是整合和分离之间随机感知变化的基础，这些变化对应于神经元活动的两个接近稳定的状态，而变化对应于两个状态之间的随机转换。此外，我们将用模型和实验来测试这两种状态，这两种状态可以通过先慢后慢的刺激坡道直接演示。 与公共健康相关：听力损失或佩戴人工耳蜗的人在嘈杂的环境中很难听到或选择一种声音。我们在听觉感知方面的研究结果将有助于阐明神经机制，使分离声音和听到不同的声音流成为可能。这些见解可能有助于开发假体，最终可以提高听力受损用户的这种能力。

DESCRIPTION (provided by applicant): My goal, as an established computational neuroscientist, is to augment my research program by acquiring new research skills and knowledge that will enable me to formulate and to answer questions relevant to auditory perception, specifically auditory streaming, and neurobiological mechanisms thereof. My career since receiving my PhD (1973) has involved computational modeling, simulation and application of nonlinear dynamical systems concepts, primarily at the cell and circuit level to identify basic mechanisms of cell and circuit dynamics: integration and input/output properties such as rhythmicity and bistability. I have dedicated most of my funded research efforts since the late-1990s to understanding issues in auditory temporal processing -- working steadily on computational modeling and (directing but not performing) in vitro patch-clamp experiments, primarily on the biophysical mechanisms of neurons in the auditory brain stem; they show exquisite ability for sub-millisecond coincidence detection for the neuronal computation of sound-localization Recently, through a collaborative research experience, I have worked with models of perceptual bistability for ambiguous visual scenes. It has been satisfying and successful and it has motivated me to acquire knowledge and skills for investigating perceptual scene issues in the auditory modality. Furthermore, while the stimuli for studying visual perceptual bistability are typically stationary, in the auditory modality, stimuli are naturally dynamic: sound is changing and therefore the structural features or cues of an "auditory object" are evolving. How we identify and track in time these features lead to compelling questions about dynamical mechanisms. I am motivated to apply dynamical systems concepts and mechanistic modeling to the streaming problem and I am enthused about investigations on the perceptual/behavioral level, a quite different level from my past research. In the long term, I plan to seek funds as a PI for recruiting, training, working with students/postdocs in auditory streaming and perception, for both modeling and psychophysical experiments. I expect to have access to an existing sound-proof acoustic chamber here at NYU for behavioral experiments. Specifically, I plan to study and work, guided by the mentorship of highly regarded auditory systems neuroscientists, to: (1) acquire foundational knowledge, literature familiarity, working communication skills, and (2) acquire the experimental know-how to design, carry-out, and supervise psychophysical experiments and computational modeling in the area of auditory streaming. My mentors will be: Dr. Elyse Sussman of Albert Einstein College of Medicine and Dr. Shihab Shamma of the University of Maryland, College Park. I will learn about design, implementation and data analysis of human psychophysics experiments for streaming, primarily, from Sussman and her staff, so that I can perform these experiments independently - including the experimental component of my research project. From Shamma, I will learn, in addition, about neurophysiological experiments on ferrets (auditory cortex) and about modeling and data analysis for his ongoing streaming research. Shamma's expertise will be valuable for the modeling component of my research project to develop mechanistic models for bistability in streaming. The process for learning background and empirical bases with both mentors will involve my reading of foundational literature, regular discussions with mentors, participating in lab meetings, and sitting-in on experiments. I plan to participate in a grad-level course at the Univ of MD in Psychoacoustics (Experimental Audiology) taught annually by Monita Chatterjee. My research strategy involves two linked components, computational modeling and experimental psychophysics, on the dynamics of auditory streaming for a potentially ambiguous stimulus. The stimulus consists of two tone-sequences of different frequencies (frequency difference, DF) and interleaved with presentation rate (PR): say, A_B_A__A_B_A__A_B_A__A_B_A__.... The subject may perceive the A- sequence and B-sequence as distinct (segregated, two streams) or as a unit (integrated, one stream) depending on the values of DF and PR; 2 streams if DF or PR is large. For long presentations, the perceptions may alternate in time (durations of a few seconds) (Pressnitzer & Hupe, 2006). My primary hypothesis is that bistability underlies the random perceptual alternations between integration and segregation and that these correspond to two near-steady states of neuronal activity and alternations correspond to random switching between the two states. Furthermore, the two states may be demonstrated directly by a slow- up-then-down stimulus ramp that we will test with models and experiments. PUBLIC HEALTH RELEVANCE: People who have hearing loss or wear cochlear implants have difficulty listening to or selecting one voice in noisy environments. The results of our work on auditory perception will help elucidate neural mechanisms that make it possible to segregate sounds and hear distinct sound streams. These insights could help with developing prosthetics that ultimately could improve this ability in hearing impaired users.