Mechanisms of multi-attribute decision-making

点击翻译按钮获取中文摘要

When making a complex decision, we often consider multiple dimensions, such as costs and qualities, that vary among choice options. Evaluating important attributes of a given option is critical for optimal choice behavior, and poor decision-making can result from an inability to properly weigh attributes, as is commonly observed in psychiatric disorders. These deficits are accompanied by alterations in the structure and function of the orbitofrontal cortex (OFC), an area critical for value-based decision-making. However, the underlying neural mechanisms and how they are disrupted remain unclear, and this limits our ability to map decision-making deficits to neural computation. The long-term goal of this proposal is to understand how the brain uses information to make optimal decisions, and our specific objective is to develop a comprehensive model of information processing in OFC during multi-attribute choices. To do this, we will use a multi-modal approach to evaluate different frameworks of decision formation. A neuroeconomics view posits that the values of different attributes are combined to compute an overall, or integrated value, and comparisons are made in the space of these option values. In contrast, other evidence suggests that direct competition between attributes, perhaps mediated by visual attention, is an important part of the decision process. Arbitrating between these models is critical to advancing theoretical frameworks that can link decision-making deficits to disordered neural computations, but a key challenge is that the steps of decision formation occur rapidly and internally, making them difficult to observe or otherwise measure. Here, we address this by combining a novel multi-attribute choice task with large-scale neural recording and population analyses necessary to reveal within-trial dynamics of otherwise covert decision-making processes. In Aim 1, we will assess how OFC codes individual attributes during multi-attribute decisions, and how this relates to classically reported integrated value signals. Next, we will assess how attention to attributes alters OFC coding, value computation, and subsequent decisions (Aim 2). Finally, in Aim 3, we propose a novel computational model of multi-attribute decisions that can determine the extent to which choices are driven by the relative values of attributes versus integrated options. Our model will also reveal latent variables that evolve during decision formation, which we will map on to neural responses. In doing so, we aim to localize specific choice processes to unique neural circuits, and also demonstrate the biological relevance of the model and its conclusions. Together, these studies leverage our combined expertise in non-human primate behavior, computational analysis, and modeling to define the neural underpinnings of multi-attribute choice in OFC. If successful, our results will not only refine the theoretical frameworks that guide decision neuroscience, but will also shed light on neural processes that underlie decision-making deficits characteristic of human psychiatric disorders.