Age effects and gaze patterns in recognising emotional expressions: An in-depth look at gaze measures and covariates

Abstract

The present study investigated predictors of age effects in emotion recognition accuracy. Older and younger adults were tested on a battery of cognitive, vision, and affective questionnaires; participants' eyes were also tracked while they completed an emotion recognition task. Older adults were worse at recognising sad, angry, and fearful expressions than younger adults. When controlling for covariates related to emotion recognition accuracy, younger adults still outperformed older adults in recognising anger and sadness. Younger adults tended to pay more attention to the eyes than older adults. Results suggest that age-related gaze patterns in emotion recognition may depend on the specific emotion being recognised and may not generalise across stimuli sets.

Keywords:

• Aging
• Emotion recognition
• Eye tracking
• Accuracy

Acknowledgements

This research was funded in part by National Institute of Aging grants (T32 AG00204, R01 AG026323). Portions of these data were presented at the 2007 Society for Personality and Social Psychology (SPSP) conference in Memphis, TN, and the 2007 Gerontological Society of America (GSA) conference in San Francisco, CA. Nora Murphy is now at the Department of Psychology, Loyola Marymount University.

We thank Corinna Löeckenhoff and Jenny Tehan Stanley for their feedback regarding an earlier version of this manuscript. We appreciate data analysis assistance from Jonathan Lehrfeld and Christina Fernandes Megias.

Notes

¹Younger adult trackable participants did not differ from non-trackable participants on cognitive, affective, and vision measures with the exception of the following: trackable younger adults (M=7.73, SD=1.26) performed better than non-trackable younger adults (M=6.78, SD=1.20) on the WAIS digit span forward, t(47) = 2.05, p<.05, d=0.60. Trackable younger adults (M=5.60, SD=1.35) performed better than non-trackable younger adults (M=4.11, SD=0.33) on the WAIS digit span backwards, t(47) = 3.25, p<.01, d=0.95. One trackable younger adult did not complete the WAIS forward and backward digit span subtests.

There was also no difference between the trackable and non-trackable older adult participants on any cognitive, affective and vision measures except for the WAIS-R digit span forward scores; trackable older adults (M=6.45, SD=1.26) performed worse than non-trackable older adults (M=7.72, SD=1.41), t(38) = − 3.00, p<.01, d=0.97. One trackable older adult did not complete the WAIS forward and backward digit span subtests.

²Paired samples t-tests tested the pairwise comparisons between the four emotion recognition scores within each age group. All pairwise comparisons were significantly different at p<.01, all ts(75) ≥ 3.00.

³There were no significant gender differences on overall emotion recognition accuracy scores or accuracy scores for each specific emotion. Thus this variable is not further discussed.

⁴Accuracy for recognising each emotion at high intensity and low intensity (3 slides per level of intensity per emotion) was also calculated. The same gaze pattern and emotion accuracy analyses were conducted within each emotion-by-intensity type (e.g., high- or low-intensity happy expression slides). While there were a few differences in terms of accuracy and gaze patterns depending on the low or high intensity of the emotion expression, these intensity accuracy patterns did not differ markedly from the overall accuracy patterns. Typically, splitting the results within an emotion by intensity reduced one or two of the correlations to a non-significant level. For brevity and clarity, we chose to focus on the overall emotion patterns and not include emotion-by-intensity results, particularly because there were no a priori predictions regarding intensity and accuracy.