Jack, Rachael E.
Cultural differences in the decoding and representation of facial expression signals.
PhD thesis, University of Glasgow.
Full text available as:
Summary. In this thesis, I will challenge one of the most fundamental assumptions of psychological science – the universality of facial expressions. I will do so by first reviewing the literature to reveal major flaws in the supporting arguments for universality. I will then present new data demonstrating how culture has shaped the decoding and transmission of facial expression signals. A summary of both sections are presented below.
Review of the Literature
To obtain a clear understanding of how the universality hypothesis developed, I will present the historical course of the emotion literature, reviewing relevant works supporting notions of a ‘universal language of emotion.’ Specifically, I will examine work on the recognition of facial expressions across cultures as it constitutes a main component of the evidence for universality. First, I will reveal that a number of ‘seminal’ works supporting the universality hypothesis are critically flawed, precluding them from further consideration. Secondly, by questioning the validity of the statistical criteria used to demonstrate ‘universal recognition,’ I will show that long-standing claims of universality are both misleading and unsubstantiated. On a related note, I will detail the creation of the ‘universal’ facial expression stimulus set (Facial Action Coding System -FACS- coded facial expressions) to reveal that it is in fact a biased, culture-specific representation of Western facial expressions of emotion. The implications for future cross-cultural work are discussed in relation to the limited FACS-coded stimulus set.
In reviewing the literature, I will reveal a latent phenomenon which has so far remained unexplained – the East Asian (EA) recognition deficit. Specifically, EA observers consistently perform significantly poorer when categorising certain ‘universal’ facial expressions compared to Western Caucasian (WC) observers – a surprisingly neglected finding given the importance of emotion communication for human social interaction. To address this neglected issue, I examined both the decoding and transmission of facial expression signals in WC and EA observers.
Experiment 1: Cultural Decoding of ‘Universal’ Facial Expressions of Emotion
To examine the decoding of ‘universal’ facial expressions across cultures, I used eye tracking technology to record the eye movements of WC and EA observers while they categorised the 6 ‘universal’ facial expressions of emotion. My behavioural results demonstrate the robustness of the phenomenon by replicating the EA recognition deficit (i.e., EA observers are significantly poorer at recognizing facial expressions of ‘fear’ and ‘disgust’). Further inspection of the data also showed that EA observers systematically miscategorise ‘fear’ as ‘surprise’ and ‘disgust’ as ‘anger.’ Using spatio-temporal analyses of fixations, I will show that WC and EA observers use culture-specific fixation strategies to decode ‘universal’ facial expressions of emotion. Specifically, while WC observers distribute fixations across the face, sampling the eyes and mouth, EA observers persistently bias fixations towards the eyes and neglect critical features, especially for facial expressions eliciting significant confusion (i.e., ‘fear,’ ‘disgust,’ and ‘anger’).
My behavioural data showed that EA observers systematically miscategorise ‘fear’ as ‘surprise’ and ‘disgust’ as ‘anger.’ Analysis of my eye movement data also showed that EA observers repetitively sample information from the eye region during facial expression decoding, particularly for those eliciting significant behavioural confusions (i.e., ‘fear,’ ‘disgust,’ and ‘anger’). To objectively examine whether the EA culture-specific fixation pattern could give rise to the reported behavioural confusions, I built a model observer that samples information from the face to categorise facial expressions. Using this model observer, I will show that the EA decoding strategy is inadequate to distinguish ‘fear’ from ‘surprise’ and ‘disgust’ from ‘anger,’ thus giving rise to the reported EA behavioural confusions. For the first time, I will reveal the origins of a latent phenomenon - the EA recognition deficit. I discuss the implications of culture-specific decoding strategies during facial expression categorization in light of current theories of cross-cultural emotion communication.
Experiment 2: Cultural Internal Representations of Facial Expressions of Emotion
In the previous two experiments, I presented data that questions the universality of facial expressions. As replicated in Experiment 1, WC and EA observers differ significantly in their recognition performance for certain ‘universal’ facial expressions. In Experiment 1, I showed culture-specific fixation patterns, demonstrating cultural differences in the predicted locations of diagnostic information. Together, these data predict cultural specificity in facial expression signals, supporting notions of cultural ‘accents’ and/or ‘dialects.’ To examine whether facial expression signals differ across cultures, I used a powerful reverse correlation (RC) technique to reveal the internal representations of the 6 ‘basic’ facial expressions of emotion in WC and EA observers. Using complementary statistical image processing techniques to examine the signal properties of each internal representation, I will directly reveal cultural specificity in the representations of the 6 ‘basic’ facial expressions of emotion. Specifically, I will show that while WC representations of facial expressions predominantly featured the eyebrows and mouth, EA representations were biased towards the eyes, as predicted by my eye movement data in Experiment 1. I will also show gaze avoidance as unique feature of the EA group.
In sum, this data shows clear cultural contrasts in facial expression signals by showing that culture shapes the internal representations of emotion.
My review of the literature will show that pivotal concepts such as ‘recognition’ and ‘universality’ are currently flawed and have misled both the interpretation of empirical work the direction of theoretical developments. Here, I will examine each concept in turn and propose more accurate criteria with which to demonstrate ‘universal recognition’ in future studies. In doing so, I will also detail possible future studies designed to address current gaps in knowledge created by use of inappropriate criteria. On a related note, having questioned the validity of FACS-coded facial expressions as ‘universal’ facial expressions, I will highlight an area for empirical development – the creation of a culturally valid facial expression stimulus set – and detail future work required to address this question. Finally, I will discuss broader areas of interest (i.e., lexical structure of emotion) which could elevate current knowledge of cross-cultural facial expression recognition and emotion communication in the future.
Actions (login required)