Dr. Ding Xiao Pan
Detecting Young Children’s Deception Using Facial Expression with High Dimensional Statistical Methods
7 March 2023 (Tuesday), 4pm
Deception is ubiquitous in young children, but previous studies have shown that it is challenging to detect children’s deception (Evans et al., 2016). To overcome this, recent studies have employed various methods to identify children’s deceptive behavior, including the use of facial expressions (Bruer et al., 2020; Zanette et al., 2016). The present study aims to investigate the potential of high-dimensional statistical methods to detect children’s deception through facial expressions. We recorded the facial expressions of 176 4-6-year-old children during the Temptation Resistance Paradigm. A machine vision algorithm (FACET) analyzed the likelihood of nine basic facial expressions. Using two parametric prediction methods (Logistic regression, linear discriminant analysis) and two non-parametric prediction methods (random forest, and adaptive boosting) with the facial expression data, we estimated the original accuracy and balanced accuracy of detecting deception. Our results showed: First, adding the slope and fitness of the curve’s shape to the mean provides a better prediction when employing a time-dependent procedure. Second, for a data set with a relatively small sample size and multiple predictor variables, a feature selection step is recommended as a part of high-dimensional statistical methods. Lastly, balanced accuracy should be used for unbalanced data.
Dr. Xiao Pan Ding has been an Assistant Professor at the Department of Psychology at the National University of Singapore since 2016. She directs NUS Child Development Lab (www.nuschildlab.com). Her research interest is children’s moral and cognitive behavior, specifically children’s honest and dishonest behavior. She uses both behavioural and neuroimaging methods to explore the cognitive and neural correlates underlying children’s deceptive behaviour.