A Recurrent Neural Network for Attenuating Non-cognitive Components of Pupil Dynamics.
A Recurrent Neural Network for Attenuating Non-cognitive Components of Pupil Dynamics.
复制标题
DOI：
10.3389/fpsyg.2021.604522
复制
发表时间：
2021
影响因子：
3.8
通讯作者：
Sajda P
中科院分区：
文献类型：
--
作者：
Koorathota S;Thakoor K;Hong L;Mao Y;Adelman P;Sajda P
人们对瞳孔动态如何反映潜在的认知过程和大脑状态越来越感兴趣。然而，问题是瞳孔的变化可能是由非认知因素引起的，例如环境中的亮度变化、适应和运动。在本文中，我们考虑如何通过建模瞳孔在现实世界环境中的反应，我们可以捕获非认知相关的变化，并去除这些变化，以提取残差信号，这是一个更好的认知和表现指标。具体来说，我们利用固定位置、持续时间、扫视和眨眼相关信息等序列测量作为深度递归神经网络（RNN）模型的输入，以预测随后的瞳孔直径。我们为一个任务建立和评估模型，在这个任务中，受试者正在观看教育视频，随后根据内容提出问题。与该任务的常用模型相比，RNN在给定序列数据的情况下预测后续瞳孔扩张的错误率最低。最重要的是，模型输出与受试者的认知表现之间的关系，这是通过观看后测试评估的。根据我们的假设，该模型捕获了非认知瞳孔动态，我们发现(1)模型的均方根误差对于表现较差的受试者小于那些在观看后测试中表现较好的受试者，(2)RNN （LSTM）模型的残差与受试者观看后测试成绩的相关性最高，(3)残差具有最高的判别性（通过ROC曲线下面积，AUC评估），用于分类高和低测试表现。与真实瞳孔大小或RNN模型预测相比。这表明，深度学习序列模型可能有助于将瞳孔反应中与亮度和适应性相关的部分与与认知和觉醒相关的部分分离开来。
There is increasing interest in how the pupil dynamics of the eye reflect underlying cognitive processes and brain states. Problematic, however, is that pupil changes can be due to non-cognitive factors, for example luminance changes in the environment, accommodation and movement. In this paper we consider how by modeling the response of the pupil in real-world environments we can capture the non-cognitive related changes and remove these to extract a residual signal which is a better index of cognition and performance. Specifically, we utilize sequence measures such as fixation position, duration, saccades, and blink-related information as inputs to a deep recurrent neural network (RNN) model for predicting subsequent pupil diameter. We build and evaluate the model for a task where subjects are watching educational videos and subsequently asked questions based on the content. Compared to commonly-used models for this task, the RNN had the lowest errors rates in predicting subsequent pupil dilation given sequence data. Most importantly was how the model output related to subjects' cognitive performance as assessed by a post-viewing test. Consistent with our hypothesis that the model captures non-cognitive pupil dynamics, we found (1) the model's root-mean square error was less for lower performing subjects than for those having better performance on the post-viewing test, (2) the residuals of the RNN (LSTM) model had the highest correlation with subject post-viewing test scores and (3) the residuals had the highest discriminability (assessed via area under the ROC curve, AUC) for classifying high and low test performers, compared to the true pupil size or the RNN model predictions. This suggests that deep learning sequence models may be good for separating components of pupil responses that are linked to luminance and accommodation from those that are linked to cognition and arousal.
登录
查看更多内容
影响因子：
4.7
作者：
Eckstein MK;Guerra-Carrillo B;Miller Singley AT;Bunge SA
通讯作者：
Bunge SA
DOI：
10.1080/13658816.2010.511718
发表时间：
2010-01-01
影响因子：
5.7
作者：
Coeltekin, A.;Fabrikant, S. I.;Lacayo, Martin
通讯作者：
Lacayo, Martin
DOI：
10.1073/pnas.1317557111
发表时间：
2014-02-04
影响因子：
11.1
作者：
de Gee, Jan Willem;Knapen, Tomas;Donner, Tobias H.
通讯作者：
Donner, Tobias H.
影响因子：
4.6
作者：
Dalrymple, Kirsten A.;Jiang, Ming;Elison, Jed T.
通讯作者：
Elison, Jed T.
影响因子：
4.6
作者：
Kirkorian HL;Anderson DR;Keen R
通讯作者：
Keen R