Prediction and learning in the infant brain: From auditory perception to social interaction

要在世界上发挥作用，就需要我们不断地预测未来，预测预期会发生什么以及何时发生，以便准备及时的最佳应对措施。例如，我们的祖先需要预测捕食者或猎物在正确的时间出现在正确的地点，以及何时何地会出现。语音的音素和音乐的音符以湍急的溪流出现，一旦发出，就消失了。将这些流实时解密并组织成有意义的单位，如单词和短语，依赖于预测和错误监测。对成年人的研究表明，大脑不断预测它预期下一步会发生什么声音，并将预测结果与实际发生的情况进行比较。预测状态和实际状态之间的不匹配会产生可以用脑电(EEG)测量的误差信号。预测误差会引起注意，并导致学习，从而将未来的预测误差降至最低。 这项拟议的研究侧重于婴儿期预测性过程的发展以及它们如何影响发展。第1部分中的研究考察了预测编码(预期的)和预测计时(在预期的时候)的发展。预测性编码将通过脑电对一系列音调中偶尔遗漏的音调的反应来衡量，比较可预测的遗漏和不可预测的遗漏。预计预测编码在出生后的第一年会有所改善。预测计时将用有节奏的序列来研究，因为它们的规律性能够预测下一个节拍的预期时间。本课程将考察大脑对节奏刺激的节奏反应的发展，以及增强的神经携带(良好的时间预测)是否会导致对声音的感知增强。其他问题包括婴儿是否对并发的听觉和触觉刺激流形成多感官预测，以及预测性学习是否通过与成年人的积极接触而得到增强。 第二部分的研究考察了母婴之间人际互动的预测。父母和婴儿之间的互动协调被认为与最佳的社会、情感和认知发展有关。在成年人中，非语言交流反映在潜意识中的动作同步以及一个人的行为对另一个人未来行为的影响。在这里，我们使用运动捕捉和数学模型来衡量父母和婴儿之间的运动协调性，以及它与压力、自主神经系统功能和早期语言发展的关系。 预测加工是知觉、认知、学习和社会发展的基础。例如，了解它在典型人群中的发展可能会导致及早发现有学习障碍风险的婴儿，并对患有产后抑郁症的母亲进行干预，这些母亲与婴儿的社交能力很差。

Functioning in the world requires that we continually predict the futurewhat is expected to happen as well as whenin order to prepare timely optimal responses. For example, our ancestors needed to predict where and when a predator or prey would go next to be in the right place at the right time. The phonemes of speech and the notes of music occur in rapid streams and, once sounded, are gone. Deciphering and organizing such streams in real time into meaningful units such as words and phrases relies on prediction and error monitoring. Studies in adults indicate that the brain continually predicts what sounds it expects next and compares its predictions with what actually happens. Mismatches between predicted and actual states generate error signals that can be measured with electroencephalography (EEG). Predictive errors engage attention and lead to learning that minimizes future prediction errors. The proposed research focuses on the development of predictive processes in infancy and how they shape development. The studies in Part 1 examine the development of predictive coding (what is expected) and predictive timing (when it is expected). Predictive coding will be measured with EEG responses to the occasional omission of tones in a sequence of tones, comparing omissions that are predictable to omissions that are not predictable. It is expected that predictive coding will improve over the first year after birth. Predictive timing will be studied with rhythmic sequences, as their regularity enables prediction of when the next beats are expected. The development of oscillatory brain responses entrained to the tempo of the rhythmic stimuli will be examined, and whether enhanced neural entrainment (good timing prediction) leads to enhanced perception the sounds. Additional questions include whether infants form multisensory predictions for concurrent streams of auditory and touch stimuli and whether predictive learning is enhanced by active engagement with an adult. The studies in Part 2 examine prediction in interpersonal interactions between mothers and infants. Interactive coordination between parents and infants is thought relate to optimal social, emotional and cognitive development. In adults, nonverbal communication is reflected in subconscious synchronization of movements and the influence of one person's actions on the future actions of the other. Here we use motion capture and mathematical models to measure movement coordination between parents and infants and how it relates to stress, autonomic nervous system function and early language development. Predictive processing is fundamental to perception, cognition, learning and social development. Understanding its development in typical populations could lead, for example, to early identification of infants at risk for learning disabilities, and interventions with mothers suffering post partum depression who are at risk for poor social interaction with their infant.