Original Article
Prioritization of emotional signals by the human auditory system: evidence from a perceptual hysteresis protocol

https://doi.org/10.1016/j.evolhumbehav.2014.07.005Get rights and content

Abstract

It is expected that natural selection has endowed our auditory apparatus with the ability to adaptively prioritize information that is crucial for survival and reproduction, such as vocal emotional signals emitted by our conspecifics, even in a noisy and dynamic natural environment (signals progressively emerge or fade away in noise as conspecifics move toward or away from us). Here, we tested the hypothesis that emotional signals are detected more easily (i.e., at lower signal-to-noise levels) and retained for a longer time (i.e., persisting in your sensory system at greater distance from the physical source) than signals bearing no emotional content, using a perceptual hysteresis protocol. Trials consisted of emotional signals (i.e., laughter and screams) or neutral signals (spectrally-rotated versions of the emotional stimuli) progressively emerging from white noise (ascending sequences) or progressively fading away in white noise (descending sequences). We demonstrated that vocal emotional signals were significantly detected at lower signal-to-noise levels than emotionally neutral signals in both ascending and descending sequences, suggesting that the human auditory system prioritizes signals bearing adaptive value.

Introduction

Emotional vocal signals are part of the behavioural repertoire of many mammal species (Hauser, 1997, Altenmüller et al., 2013), including humans (Sauter and Eimer, 2010, Sauter et al., 2010, Scherer et al., 2003), and are known to play a prominent role during social interactions. Among these vocalizations, screams and laughter, calls that signal distinct adaptive challenges in the environment (respectively, threat-related or agonistic contexts, and rewarding, social-bonding or playful interactions) seem to have been preserved throughout phylogeny (Morton, 1977, Vettin and Todt, 2005, Davila Ross et al., 2009). It is therefore expected that mechanisms favouring the production of attuned screams and laughter bouts (so as to produce signals that are effective in selectively affecting recipients) have co-evolved with mechanisms allowing for the accurate perception of these signals so that perceivers would be able to prioritize auditory information that is crucial for adaptation, particularly in noisy and dynamic environments, where signals are progressively emerging from or fading away in noise.

So far, research on the detection of vocal emotional signals (as well as of visual emotional signals) has primarily focused on the detection of static stimuli (Scherer et al., 2011, Sacharin et al., 2012), that is to say, on the detection of stimuli for which the amplitude does not evolve over time (Calvo and Esteves, 2005, Hock et al., 1993). In addition, vocal emotions detection has seldom been studied under noisy conditions (Schuller et al., 2007, Tawari and Trivedi, 2010). More crucially, the detection of signals whose amplitude evolves over time in noise has received even less consideration. While studies addressing emotion detection with static stimuli mirror situations where the emitting source and the observer are fixed in space in a non-noisy environment, the present protocol mirrors more realistic or ecological situations where sources and observers are moving in a noisy environment: a source moving in a noisy space away from (or toward) an observer would lead to a progressive decrease (or increase) of signal-to-noise ratio (hereafter, SNR).

In this research, we tested the hypothesis that the auditory sensory system prioritizes vocal emotional signals of fear (scream) and amusement (laughter) when they are presented within a noisy dynamic setting. We predicted that emotional signals would be detected more easily (i.e., at lower signal-to-noise ratio) and maintained for longer (i.e., at lower signal-to-noise ratio too) than neutral signals. To test our predictions, we used an auditory protocol of hysteresis which provides a very effective way to test the “persistence of a percept despite parameter change to values favouring the alternative pattern” (Hock et al., 1993, p. 63).

More precisely, hysteresis shows that the content of one's perception at time t depends on the recent history of the perceptual system. In previous research, hysteresis effects have been shown to occur in many contexts: bistability (Gepshtein and Kubovy, 2005, Hock et al., 1993, Hock et al., 1997, Schwiedrzik et al., 2012), form perception (Large, Aldcroft, & Vilis, 2005), letter recognition (Kleinschmidt, Büchel, Hutton, Friston, & Frackowiak, 2002), stereopsis and binocular rivalry (Buckthought, Kim, & Wilson, 2008), sentences (Rączaszek, Tuller, Shapiro, Case, & Kelso, 1999), speech categorization (Tuller, Case, Ding, & Kelso, 1994) and facial emotions (Sacharin et al., 2012). Of particular interest here, Sacharin et al. (2012) showed that when subjects are presented with certain facial emotional expressions evolving over time from a particular emotion to another, they persist in perceiving the original emotion. For instance, when presented with faces evolving from the expression of anger to that of disgust, and from disgust to anger in a subsequent trial, the threshold at which subjects stop reporting seeing anger is lower in the anger-to-disgust trials, than the threshold at which they report starting perceiving anger in disgust-to-anger trials.

Hysteresis is usually investigated using designs comprising "ascending" and "descending" sequences, that is, sequences ordered in terms of a certain physical parameter. Here, ascending and descending sequences consisted of many steps with different signal-to-noise ratios between a target and a mask. The masks consisted of bursts of white noise of constant intensity. The targets were short emotional or neutral auditory signals. The SNR was progressively increased in ascending sequences or progressively decreased in descending sequences. A similar methodology has been used in previous experiments, notably in the visual domain (e.g., Kleinschmidt et al., 2002), and is known to efficiently produce hysteresis effects in subjects.

The use of this methodology revealed, in our experiment, significant greater detection at lower SNR-levels for emotional signals compared to neutral ones, in both ascending and descending sequences, suggesting that the human auditory system can prioritize signals bearing adaptive value.

Section snippets

Participants

Eight participants (7 females, mean age of 25.25 years +/− 0.619 SEM) participated in the study after having given their informed consent. None of our participants reported history of hearing problems, and all were naive regarding the purpose of the experiment. All participants were recruited from the database of the ‘Relais d’Information sur les Sciences de la Cognition’ (RISC, Paris, France). They received a compensation of 50€ for their participation.

Experimental setup

The experiment was conducted in a quiet

Results

Fig. 2 shows the SNR-value for which the different categories of stimuli reach (ascending sequences) and drop below (descending sequences) 75% of detection. We directly analysed differences between emotions and neutral stimuli in terms of detection (at which thresholds stimuli reach 75% of detection for ascending sequences) and persistence (at which thresholds stimuli drop below 75% of detection for descending sequences) (see Fig. 2). The ANOVA revealed main effects of Direction (F1,7 = 34.728, p =

Discussion

By designing an experiment in which the amplitude of specific signals evolved monotonically over time in noise, we revealed that emotional signals were detected more easily than neutral signals and maintained for longer over noise compared to neutral signals. Indeed, they were detected at lower signal-to-noise ratios in both ascending and descending sequences compared to neutral stimuli. The present findings support the idea that, within a noisy dynamic setting, the auditory system prioritizes

Supplementary Material

The following is the supplementary data to this article.

Simulated dB-values (with the sigmoid function) at which each participant reaches 75% of detection (in ascending trials) and drops below 75% of detection (in descending trials).

Acknowledgments

The authors would like to thank Daniel Pressnitzer for technical assistance and for important comments about the manuscript, Disa Sauter for sharing the stimuli, Valentin Wyart for analytics tools, Victor Benichoux, Hadrien Orvoën, Elisabeth Pacherie, Ariadna Fernandez for helpful comments. The research was supported by an ANR-11-0001-02 PSL*, an ANR-10-LABX-0087, a ED3C/UPMC scholarship (J.R.M.) and a DGA-MRIS scholarship (G.D.).

References (43)

  • M. Davila-Ross et al.

    Aping expressions? Chimpanzees produce distinct laugh types when responding to laughter of others

    Emotion

    (2011)
  • R.I.M. Dunbar et al.

    Social laughter is correlated with an elevated pain threshold

    Proceedings of the Royal Society B: Biological Sciences

    (2012)
  • G.T. Fechner

    Elemente der Psychophysik (Elements of Psychophysics).

    Trans. Adler, H. E. Holt, Rinehart and Winston, New York, USA

    (1860)
  • S. Gepshtein et al.

    Stability and change in perception: Spatial organization in temporal context

    Experimental Brain Research

    (2005)
  • K. Grammer

    Strangers meet: Laughter and nonverbal signs of interest in opposite-sex encounters

    Journal of Nonverbal Behavior

    (1990)
  • T. Green et al.

    Adaptation to spectrally-rotated speech

    The Journal of the Acoustical Society of America

    (2013)
  • J.A. Hall

    Gender effects in decoding nonverbal cues

    Psychological Bulletin

    (1978)
  • J.A. Hall

    Nonverbal sex differences: Accuracy of communication and expressive style (reprint.)

    (1990)
  • M.D. Hauser

    The evolution of communication

    (1997)
  • H.S. Hock et al.

    Bistability and hysteresis in the organization of apparent motion patterns

    Journal of Experimental Psychology: Human Perception and Performance

    (1993)
  • H.S. Hock et al.

    Dynamic, state-dependent thresholds for the perception of single-element apparent motion: Bistability from local cooperativity

    Attention, Perception, & Psychophysics

    (1997)
  • Cited by (2)

    • Dynamic emotion transitions based on emotion hysteresis

      2017, 2017 7th International Conference on Affective Computing and Intelligent Interaction, ACII 2017
    View full text