Your gaze betrays your age

Visual attention networks are so pervasive in the human brain that eye movements carry a wealth of information that can be exploited for many purposes. In this paper, we present evidence that information derived from observers' gaze can be used to infer their age. This is the first study showing that simple features extracted from the ordered sequence of fixations and saccades allow us to predict the age of an observer. Eye movements of 101 participants split into 4 age groups (adults, 6–10 year-old, 4–6 year-old and 2 year-old) were recorded while exploring static images. The analysis of observers' gaze provides evidence of age-related differences in viewing patterns. Therefore, we extract from the scanpaths several features, including fixation durations and saccade amplitudes, and learn a direct mapping from those features to age using Gentle AdaBoost classifiers. Experimental results show that the proposed image-blind method succeeds in predicting the age of the observer up to 92% of the time. The use of predicted salience does not further improve the classification's accuracy.

[1]  Antoine Coutrot,et al.  Computational Model for Predicting Visual Fixations from Childhood to Adulthood , 2017, ArXiv.

[2]  Esa Rahtu,et al.  Stochastic bottom-up fixation prediction and saccade generation , 2013, Image Vis. Comput..

[3]  Zhi Liu,et al.  Saccadic model of eye movements for free-viewing condition , 2015, Vision Research.

[4]  Stephen V. Stehman,et al.  Selecting and interpreting measures of thematic classification accuracy , 1997 .

[5]  Eva Aring,et al.  Visual fixation development in children , 2007, Graefe's Archive for Clinical and Experimental Ophthalmology.

[6]  John M. Henderson,et al.  Predicting Cognitive State from Eye Movements , 2013, PloS one.

[7]  David S Wooding,et al.  Eye movements of large populations: II. Deriving regions of interest, coverage, and similarity using fixation maps , 2002, Behavior research methods, instruments, & computers : a journal of the Psychonomic Society, Inc.

[8]  L. Itti.  New Eye-Tracking Techniques May Revolutionize Mental Health Screening , 2015, Neuron.

[9]  Yoav Freund,et al.  A decision-theoretic generalization of on-line learning and an application to boosting , 1995, EuroCOLT.

[10]  S Ullman,et al.  Shifts in selective visual attention: towards the underlying neural circuitry. , 1985, Human neurobiology.

[11]  O. Meur,et al.  Introducing context-dependent and spatially-variant viewing biases in saccadic models , 2016, Vision Research.

[12]  Thierry Baccino,et al.  Methods for comparing scanpaths and saliency maps: strengths and weaknesses , 2012, Behavior Research Methods.

[13]  Nicolas Riche,et al.  RARE2012: A multi-scale rarity-based saliency detection with its comparative statistical analysis , 2013, Signal Process. Image Commun..

[14]  Yoav Freund,et al.  A decision-theoretic generalization of on-line learning and an application to boosting , 1997, EuroCOLT.

[15]  W. Einhäuser,et al.  Effects of aging on eye movements in the real world , 2015, Front. Hum. Neurosci..

[16]  Antoine Coutrot,et al.  Scanpath modeling and classification with hidden Markov models , 2017, Behavior Research Methods.

[17]  I Visser,et al.  Do infants have the horizontal bias? , 2016, Infant behavior & development.

[18]  K. Velanova,et al.  Development of eye-movement control , 2008, Brain and Cognition.

[19]  Fernando Vilariño,et al.  Low Cost Eye Tracking: The Current Panorama , 2016, Comput. Intell. Neurosci..

[20]  Antoine Coutrot,et al.  Age-dependent saccadic models for predicting eye movements , 2017, 2017 IEEE International Conference on Image Processing (ICIP).

[21]  Sebastian Pannasch,et al.  The maturation of eye movement behavior: Scene viewing characteristics in children and adults , 2014, Vision Research.

[22]  Antoine Coutrot,et al.  Face exploration dynamics differentiate men and women. , 2016, Journal of vision.