Institute of Cognitive Neuroscience, University College London, London WC1 3AR, UK.
Institute of Cognitive Neuroscience, University College London, London WC1 3AR, UK.
Neuroimage. 2021 May 15;232:117863. doi: 10.1016/j.neuroimage.2021.117863. Epub 2021 Feb 19.
To interact meaningfully with its environment, an agent must integrate external information with its own internal states. However, information about the environment is often noisy. In this study, we identify a neural correlate that tracks how asymmetries between competing alternatives evolve over the course of a decision. In our task participants had to monitor a stream of discrete visual stimuli over time and decide whether or not to act, on the basis of either strong or ambiguous evidence. We found that the classic P3 event-related potential evoked by sequential evidence items tracked decision-making processes and predicted participants' categorical choices on a single trial level, both when evidence was strong and when it was ambiguous. The P3 amplitudes in response to evidence supporting the eventually selected option increased over trial time as decisions evolved, being maximally different from the P3 amplitudes evoked by competing evidence at the time of decision. Computational modelling showed that both the neural dynamics and behavioural primacy and recency effects can be explained by a combination of (a) competition between mutually inhibiting accumulators for the two categorical choice outcomes, and (b) a context-dependant urgency signal. In conditions where evidence was presented at a low rate, urgency increased faster than in conditions when evidence was very frequent. We also found that the readiness potential, a classic marker of endogenously initiated actions, was observed preceding movements in all conditions - even when those were strongly driven by external evidence.
为了与环境进行有意义的交互,智能体必须将外部信息与其内部状态集成起来。然而,环境信息通常是嘈杂的。在这项研究中,我们确定了一个神经相关物,它可以跟踪竞争选项之间的不对称性如何在决策过程中演变。在我们的任务中,参与者必须随着时间的推移监测离散视觉刺激的流,并根据强或模糊的证据来决定是否采取行动。我们发现,经典的 P3 事件相关电位被序列证据项目引发,它跟踪决策过程,并在单个试验水平上预测参与者的分类选择,无论是在证据强的时候还是在证据模糊的时候。对支持最终选择选项的证据的 P3 振幅随着决策的演变而随时间增加,在决策时与竞争证据的 P3 振幅最大不同。计算模型表明,神经动力学和行为的首要和最近效应都可以通过以下两种因素的结合来解释:(a)两个分类选择结果之间相互抑制累加器之间的竞争,以及(b)依赖上下文的紧急信号。在证据以低速率呈现的情况下,紧急情况的增加速度比在证据非常频繁的情况下更快。我们还发现,准备电位是内源性发起动作的经典标志物,在所有条件下都观察到了准备电位,即使这些条件都是由外部证据强烈驱动的。