Perception and Action: Ideal Observers and Actors

描述（由申请人提供）：人类在决策或行动的所有方面都可能是随机的情况下做出决策并执行行动。基于感官信息的行动计划有五个组成部分。首先，受试者具有关于环境状态的先验信息，包括当前位置和速度 附近的物体和主体自己的身体，这可以概括为可能的世界状态的概率分布。第二，受试者具有关于环境的当前状态的感觉输入，由于物理和神经噪声，这是不确定的。第三，将这两个信息源结合起来以决定预期的动作（按下按钮、手臂或眼睛运动，或包括对潜在后续感官输入的反应的复杂计划）。第四，由于电机噪声，产生的动作可能与预期的动作不同。最后，结果行为与当前环境的相互作用会导致一个结果（损失或收益），而这个结果也可能是不确定的。由于所有这些随机成分，视觉任务和运动规划需要计算，相当于风险下的决策。在我们最近的工作中，我们已经证明了人类在视觉任务中几乎是最优的，因为他们最大化了预期收益，以及人类行为次优的其他情况。 我们提出的实验，以更好地了解人类行为的性质，在视觉和视觉任务。我们经常使用具有实验者指定的奖励/惩罚结构的任务;这种新颖的方法允许我们将行为与最大化预期收益的最优策略进行比较。我们提出以下问题，并提出实验来解决每个问题。(1)在视觉引导的运动中，行为是如何计划的？我们将研究用于编码视觉引导动作的坐标系统，以及运动编码如何影响视觉系统适应不断变化的条件的能力。(2)人类在视觉搜索任务中的表现，明确定义了收益和损失，告诉我们关于视觉模式和视觉不确定性的编码是什么？我们将比较人类在视觉搜索任务中的表现，理想的观察者模型，最大限度地提高预期收益的情况下，不对称的回报。这些实验的结果将使我们能够区分不同的假设编码的视觉信息在周边。在这两个目标中，我们使用视觉表现的模式（同时执行到达，扫视或按键）来了解视觉刺激，不确定性和视觉引导运动的潜在编码。这些研究将阐明视觉刺激和运动的编码方式，以及视觉如何用于指导行动。

DESCRIPTION (provided by applicant): Humans make decisions and perform actions in situations in which all aspects of the decision or action are potentially stochastic. There are fiv components to the planning of an action based on sensory information. First, the subject has prior information about the state of the environment including the current positions and velocities of nearby objects and of the subject's own body, which can be summarized as a probability distribution across possible world states. Second, the subject has sensory input about the current state of the environment, which is uncertain due to physical and neural noise. Third, these two sources of information are combined to decide on an intended action (button press, arm or eye movement, or a complex plan that includes responses to potential subsequent sensory inputs). Fourth, the resulting action can differ from the intended one due to motor noise. Finally, the interaction of the resulting action with the current environment leads to a consequence (a loss or gain), and this consequence may be uncertain as well. As a result of all these stochastic components, visual tasks and movement planning require a calculation that is equivalent to decision-making under risk. In our recent work, we have demonstrated that humans are nearly optimal in visuomotor tasks in that they maximize expected gain, and other circumstances in which human behavior is suboptimal. We propose experiments to better understand the nature of human behavior in visual and visuomotor tasks. We often use tasks with an experimenter-specified reward/penalty structure; this novel approach allows us to compare behavior with the optimal strategy that maximizes expected gain. We ask the following questions and propose experiments to address each. (1) How is behavior planned in visually guided movements? We will investigate the coordinate systems used to encode visually guided actions and how the encoding of movements affects the ability of the visuomotor system to adapt to changing conditions. (2) What does human performance in visual search tasks with clearly defined gains and losses tell us about the encoding of visual patterns and visual uncertainty? We will compare human performance in visual search tasks to ideal-observer models that maximize expected gain in situations with asymmetric payoffs. The results of these experiments will enable us to distinguish different hypotheses about the encoding of visual information in the periphery. In both aims we use patterns of visuomotor performance (while performing a reach, saccade, or keypress) to learn about the underlying encoding of visual stimuli, uncertainty, and visually guided movement. These studies will shed light on the way in which visual stimuli and movements are encoded, and on how vision is used to guide action.