Behavior Research Methods (2018) 50:1415–1429

https://doi.org/10.3758/s13428-018-1027-6

Affective auditory stimulus database: An expanded version

of the International Affective Digitized Sounds (IADS-E)
Wanlu Yang 1 & Kai Makita 2,3 & Takashi Nakao 1 & Noriaki Kanayama 2 & Maro G. Machizawa 2 & Takafumi Sasaoka 2 &
Ayako Sugata 4,5 & Ryota Kobayashi 1 & Ryosuke Hiramoto 1 & Shigeto Yamawaki 2 & Makoto Iwanaga 5 & Makoto Miyatani 1

Published online: 8 March 2018

# Psychonomic Society, Inc. 2018

Abstract
Using appropriate stimuli to evoke emotions is especially important for researching emotion. Psychologists have provided several
standardized affective stimulus databases—such as the International Affective Picture System (IAPS) and the Nencki Affective
Picture System (NAPS) as visual stimulus databases, as well as the International Affective Digitized Sounds (IADS) and the
Montreal Affective Voices as auditory stimulus databases for emotional experiments. However, considering the limitations of the
existing auditory stimulus database studies, research using auditory stimuli is relatively limited compared with the studies using
visual stimuli. First, the number of sample sounds is limited, making it difficult to equate across emotional conditions and
semantic categories. Second, some artificially created materials (music or human voice) may fail to accurately drive the intended
emotional processes. Our principal aim was to expand existing auditory affective sample database to sufficiently cover natural
sounds. We asked 207 participants to rate 935 sounds (including the sounds from the IADS-2) using the Self-Assessment Manikin
(SAM) and three basic-emotion rating scales. The results showed that emotions in sounds can be distinguished on the affective
rating scales, and the stability of the evaluations of sounds revealed that we have successfully provided a larger corpus of natural,
emotionally evocative auditory stimuli, covering a wide range of semantic categories. Our expanded, standardized sound sample
database may promote a wide range of research in auditory systems and the possible interactions with other sensory modalities,
encouraging direct reliable comparisons of outcomes from different researchers in the field of psychology.

Keywords Emotion . Affective auditory stimuli . Affective ratings . International Affective Digitized Sounds . SAM

Emotion, as one of the most significant issues in psychology

and psychophysiology, has been thoroughly examined in the
Electronic supplementary material The online version of this article field of cognitive psychology (Lindquist, 2010). For a com-
(https://doi.org/10.3758/s13428-018-1027-6) contains supplementary plete understanding of emotional processes, it is especially
material, which is available to authorized users.
important to select appropriate experimental materials that
complement the experimental purposes appropriately (for a
* Wanlu Yang
yangwanlu0710@gmail.com review, see Gerdes, Wieser, & Alpers, 2014). In daily human
life, the various types of external stimulus that we receive
1 across different modalities not only have powerful effects on
Department of Psychology, Graduate School of Education,
Hiroshima University, 1-1-1, Kagamiyama, subjective senses and evoked emotions (e.g., Fazio, 2001;
Higashi-Hiroshima 739-0046, Japan Jaquet, Danuser, & Gomez, 2012; Marin, Gingras, &
2
Department of Psychiatry and Neurosciences, Graduate School of Bhattacharya, 2012) but also influence our decision-making
Biomedical & Health Sciences, Hiroshima University, and subsequent behavior (mood management theory;
Hiroshima, Japan Zillmann, 1988). Therefore, many standardized affective stim-
3
Research Center for Child Mental Development, University of Fukui, ulus databases were provided and have been extensively used
Fukui, Japan in studies on emotion.
4
Ogaki Women’s College, Gifu, Japan The emotional processes have been intensively studied par-
5
Graduate School of Integrated Arts and Sciences, Hiroshima ticularly in the visual system. In many psychological and neu-
University, Hiroshima, Japan roimaging studies, visual stimuli have been used to evoke
1416 Behav Res (2018) 50:1415–1429

certain types of emotion such as affective pictures (e.g., according to the Self-Assessment Manikin (SAM; Bradley
Buodo et al., 2017; Lang, Greenwald, Bradley, & Hamm, & Lang, 1994).
1993), emotional facial expressions (e.g., Esslen, Pascual- However, previous studies have exposed that IADS-2, as a
Marqui, Hell, Kochi, & Lehmann, 2004; Shuman, Clark- standardized international affective stimulus database, had
Polner, Meuleman, Sander, & Scherer, 2015), and written af- limitations and deficiencies (Choi et al., 2016; da Silva &
fective words (e.g., Schindler & Kissler, 2016). Among other Backs, 2015). First, as compared to most visual stimulus da-
senses (i.e., audition, touch, taste, and olfaction), auditory tabases, the number of stimuli in IADS-2 is insufficient (only
stimuli also profoundly provoke human emotions (Redondo, 167 sounds; the International Affective Picture System [IAPS]
Fraga, Padrón, & Piñeiro, 2008). However, research on audi- has 1,182 pictures: Lang, Bradley, & Cuthbert, 1997, 2008;
tory system is much less frequent than on visual system the Affective Norms for English Words [ANEW] has 1,034
(Gerdes et al., 2014). One of the potential reasons would be words: Bradley & Lang, 2007b). The small number of sound
that visual stimuli are easier to manipulate and control as com- stimuli means that the distribution of IADS-2 in the
pared with auditory stimuli (Bradley & Lang, 2000). For an bidimensional affective space of valence and arousal is un-
instance, in contrast with static pictures of facial expressions even as compared to IAPS (see Fig. 1a and b, respectively).
or written words, auditory stimuli are dynamic over time We counted the number of sounds for the IADS-2 in each
(Banse & Scherer, 1996; Droit-Volet, Ramos, Bueno, & quadrant of arousal–valence space, with the origin at ratings
Bigand, 2013; Juslin & Laukka, 2003), requiring continuous of 5. Consequently, the numbers of sounds from the first quad-
perception, integration, and interpretation for recognition rant to the fourth quadrant are 50, 31, 16, and 70, respectively,
(Schirmer & Kotz, 2006). whereas the numbers of pictures for the IAPS in each quadrant
In recent decades, researches on emotion using auditory are 225, 420, 225, and 312, respectively. The biased distribu-
stimuli have gradually increased following the advancement tion of sounds makes it difficult to choose the same number of
and growing popularity of digital information storage and sounds from different emotional conditions, and it is extreme-
software to conveniently store and manipulate digitized data; ly difficult to keep experimental conditions balanced like they
this has made experimental control of auditory stimuli increas- can be with the IAPS. Second, the number of sounds in each
ingly feasible (Bradley & Lang, 2000). For instance, Fabiani, semantic category is varied: animals (N = 18), people (N =
Kazmerski, Cycowicz, and Friedman (1996) developed a 52), nature (N = 7), daily routine sounds (N = 28), transport (N
large battery of brief (400 ms), seemingly neutral auditory = 15), electric sounds (N = 5), breaking sounds (N = 4), music
stimuli (i.e., animals, noises, artificial sounds, etc.) to investi- (N = 12), and scenarios (N = 26). The classification was eval-
gate novelty effects in the event-related potential oddball par- uated by ten participants in a pilot study. Considering the
adigm. Czigler, Cox, Gyimesi, and Horváth (2007) had irrelevant number of sounds, it is also difficult to select exper-
attempted to investigate brain processes related to unpleasant imental materials from each semantic category equally.
auditory stimuli. Armony, Aubé, Angulo-Perkins, Peretz, and In addition to the IADS, other sets of auditory stimulus
Concha (2015) performed an fMRI experiment to compare the databases, including the Montreal Affective Voices database
neural activities evoked by music and voice. Nevertheless, (Belin, Fillion-Bilodeau, & Gosselin, 2008), musical excerpts
there are flaws in these experimental materials. Most of the (Vieillard et al., 2008), Portuguese sentences and
auditory stimuli used in emotional research were selected by pseudosentences for research on emotional prosody (Castro
researchers a priori or from a limited number of participants’ & Lima, 2010), and vocal emotional stimuli in Mandarin
ratings in a preliminary experiment. Such a limited number Chinese (Liu & Pell, 2012) have also been developed.
and type of stimuli make it difficult to compare the outcomes However, the materials in these databases are limited to music
and to replicate them across research labs (Bradley & Lang, pieces or recorded vocalizations. Studies that are based on
2007a). these artificial databases may fail to accurately predict the
To address these problems and encourage research human emotion process, because the sounds we hear in daily
assessing basic and applied problems in psychological sci- life are much more extensive. Therefore, inventing a standard-
ence, the NIMH Center for Emotion and Attention at the ized natural emotional auditory stimulus database that con-
University of Florida has developed a standardized auditory tains sufficient stimuli for emotional research is urgently re-
stimulus database named the International Affective Digitized quired (Gerdes et al., 2014).
Sounds (Bradley & Lang, 1999 [IADS-1]; Bradley & Lang, In this study, we aimed to expand the IADS-2 (Bradley &
2007a [IADS-2]). The latest version, IADS-2 (Bradley & Lang, 2007a) and provide a high-quality standardized, emo-
Lang, 2007a), consists of 167 digitally recorded natural tionally evocative auditory stimuli database with contents
sounds that are common in daily life, such as a baby laughing, from across a wider range of semantic categories. We based
someone cooking, a thunderstorm, erotic sounds, and so on. our database on the IADS-2 because it is the only auditory
Each sound lasts 6 s and is characterized by the affective database that includes a wide range of semantic categories
dimensions of valence, arousal, and dominance or control and, as compared with other auditory databases, is also the
Behav Res (2018) 50:1415–1429 1417

(Ekman, 1992). However, because the number of sounds par-

ticipants needed to examine was enormous in this study, we
focused only on a subset of the six potential emotions to be
rated. We selected happiness, sadness, and fear, for these can
be distinguished on the dimensions of valence and arousal,
and have been studied well in many psychological research
(Vieillard et al., 2008). Moreover, according to Pell and Kotz
(2011), sadness and fear can be recognized more accurately in
auditory stimuli than other basic emotions, while the happi-
ness is the opposite of sadness (Vieillard et al., 2008), and
therefore we considered the three basic emotions is suitable
for our study. By omitting other emotions to be rated, we
aimed to minimize fatigue that could result from a too-long
experimental time.
Moreover, although the ratings for IADS-2 were collected
from a group, we improved on this procedure by designing an
individual experiment in a soundproofing room, to avoid in-
terference between the participants and noise. The affective
ratings measured in the individual experiment would be better
adapted to laboratory experiments, because most psychologi-
cal laboratory experiments are performed individually, and
emotions evoked in the same situation could be reexperienced
more precisely.

Method

Participants

Two hundred seven Japanese participants (104 females and

103 males, mean age = 21.32 years, SD = 2.38 years) from
Hiroshima University (except the one who had just graduated
from high school) were recruited. Over the time course of the
study, the data were collected in two different rooms, where
exactly the same procedures were carried out at the two loca-
tions except for geographical differences between the sites.
We report the details for completeness. None of the partici-
pants had any history of neurological or psychiatric disorder,
Fig. 1 Distributions of overall mean values for the 167 sounds of the and 93% of them were right-handed. All experimental proto-
IADS-2 (a) and the visual stimuli of the IAPS (b) in the valence and cols were conducted in accordance with guidelines approved
arousal affective dimensions by the Ethical Committee of the Graduate School of
Education, Hiroshima University. Written informed consent
one most frequently used in emotional research (Gerdes et al., was obtained from each participant, and monetary rewards
2014; Marchewka, Zurawski, Jednoróg, & Grabowska, 2014). were given to all participants as compensation for their
We termed the new auditory stimulus database IADS-E, an participation.
expanded version of the IADS-2. The IADS-E is freely acces-
sible to the scientific community for noncommercial use by Materials
request to the corresponding author.
For maintaining consistency in theory, our study also used Sound stimuli
the SAM (Bradley & Lang, 1994) as the IADS-2 did.
Furthermore, to provide a way to conduct more controlled (1) Sound collection: Some of the sounds were assembled
and refined experiments, depending on the experimental ob- from the Internet and sampled using computer music
jective, we tried to include the evaluation for basic emotions software, including various sounds heard in daily life
1418 Behav Res (2018) 50:1415–1429

(e.g., rain, writing, laughing, and barking), and some automation. The software, running on a personal comput-
were composed by a composer. Depending on the dura- er, controlled the presentation of instructions on the com-
tion of the sound (sounds less than 1.5 s were excluded), puter display as well as the presentation of sounds via a
naturalness, and the emotional ratings from a preliminary speaker (GX-70HD II, ONKYO Corp., Japan). It also
experiment, after careful filtering, an initial observation registered the ratings that participants entered using the
revealed 1,226 sounds, from which 768 sounds were keyboard.
selected.

We manipulated the 768 newly collected sounds to 6 s with Emotional rating measurements
the SoundEngine Free Version 5.21 software (Coderium.,
Inc., Japan) as this is the IADS-2’s standard duration. (1) Self-Assessment Manikin (SAM): This system for non-
Because differences in physical properties may influence verbal pictorial assessment, devised by Lang (1980)
emotional ratings (Lolli, Lewenstein, Basurto, Winnik, & (also seen in Bradley & Lang, 1994) to measure
Loui, 2015), we edited three physical properties (frequency, emotional ratings in the three dimensions of valence,
44.1 kHz; bit rate, 1,411 kbps; and channels, two) equally for arousal, and dominance, was used. The SAM scale
all sounds (including those from IADS-2). we used contained five affective graphical figures,
with a space between each two consecutive pictures,
(2) Experimental materials: When the new sound stimuli defining a 9-point scale for each dimension (Fig. 2):
were added to those from the IADS-2, we had collected valence (ranging from 1, depicted by a frowning,
935 sound stimuli for this experiment. The sounds were unhappy figure, to 9, a smiling, happy figure), arous-
preliminarily divided into ten semantic categories—ani- al (ranging from 1, depicted by a relaxed, sleepy
mals (N = 54), people (N = 74), nature (N = 70), daily figure, to 9, an excited, wide-eyed figure), and dom-
routine sounds (N = 187), transport (N = 59), electric inance (ranging from 1, depicted by a small figure
sounds (N = 64), sound effects (N = 171), breaking representing being controlled, to 9, a large figure
sounds (N = 56), music (N = 170), and scenarios (N = representing being in control).
30)—by ten participants who did not participate in the (2) Basic emotion rating scales: Three basic emotion rating
main experiment. Each sound’s physical property values scales—happiness, sadness, and fear—were presented as
were calculated using the Amadeus Pro sound editing supplementary evaluations. The ratings were measured
version 2.2.2 software (HairerSoft., Inc. UK), shown in from 1 point, for not at all, to 9 points, for extremely, and
the fourth sheet of the supplementary materials were shown by words. Both the SAM scales and the
(Sounds2nd). The peak sound intensity at presentation basic emotion rating scales were presented in a pseudo-
ranged from 50.4 to 88 dB(A), as measured using a NL- random order, to control for order effects (Castonguay,
42 sound level meter (RION Co., Ltd., Japan); the max- Sabiston, Crocker, & Mack, 2014; Krosnick & Presser,
imum level (Lmax), the minimum level (Lmin), the equiv- 2010). All ratings were further analyzed using the statis-
alent continuous level (Leq), and the exposure level (LE) tical package IBM SPSS Version 20.0 software (IBM
are shown on the fifth sheet of the supplementary mate- Corp., US).
rials (Sounds3rd). All sounds were presented with a
prompt fixed volume and generally varied according to
the respective natural volumes in the environment.

To ensure that the participants focused on the test, and to

avoid fatigue, we distributed the 935 sounds across ten cate-
gories into nine groups of 103–104 sounds. Each group was
divided into two blocks (each block contained 52 sounds,
except one that contained 51 sounds). Each participant rated
only one of the nine groups, and each sound was rated by at
least 22 participants. The sounds in each block were presented
in a predefined order, such that no more than three sounds that
correspond to same semantic category were presented
consecutively.
The experimental procedure was programmed using the
Presentation Version 18.1 Build 03.31.15 software Fig. 2 Self-Assessment Manikins (SAMs) for valence (top), arousal
(Neurobehavioral Systems, Inc., USA) for process (middle), and dominance (bottom).
Behav Res (2018) 50:1415–1429 1419

Measurement of mental state conducted at the participant’s own pace. The average duration
of one trial was approximately 30 s. Participants were made to
To identify the participants’ psychological states when they rate all 52 sounds in one block at a stretch. The order of the two
participated in the experiment, the Japanese version (Shimizu blocks was counterbalanced among the participants who rated the
& Imae, 1981) of the State–Trait Anxiety Inventory (STAI; sounds in the same group. A 5-min break was inserted between
Spielberger, Gorsuch, & Lushene, 1970) was used. This ques- the two experimental blocks to reduce the participants’ fatigue.
tionnaire is used to diagnose and distinguish a participant’s Each experimental session lasted approximately 1 h 10 min.
level of anxiety at the time of filling out the questionnaire.
The STAI has two parts: state anxiety and trait anxiety. Our
study used the state anxiety part (STAI-S), which consists of Results and discussion
20 statements that measure anxiety according to the situation.
All items are rated on a four-point scale from almost never to Reliability
almost always. A higher score indicates a higher level of
anxiety. We calculated the STAI-S scores for all participants. The
means and standard deviations of the STAI-S scales by gender
Procedure were 40.20 (SD = 5.57) for female participants and 41.34 (SD
= 6.28) for male participants. Mean scores of STAI-S in this
The participants were guided into a soundproof experimental study were in the normal range (32–42 points for male partic-
booth and sat in front of a computer monitor at a distance of ipants and 31–41 points for female participants) of the mental
approximately 70 cm. After receiving a detailed briefing on state (Iwata & Higuchi, 2000). These results suggested that
the experimental purpose and providing informed consent, none of the participants showed extreme uneasiness on the
two online pre-questionnaire investigations, the STAI-S day of the experiment.
(Spielberger et al., 1970), and the Edinburgh Handedness Ratings for each sound of the IADS-E are presented in the
Inventory (Oldfield, 1971), were administered. The investiga- first sheet to third sheet of the supplementary materials. The
tions lasted approximately 10 min. supplementary materials list the mean values (M) and standard
On the affective rating task, first of all, six practice sounds deviations (SD) for arousal (Aro), valence (Val), dominance
that were excluded from the experimental stimuli but had the (Dom), fear (Fear), happiness (Hap), and sadness (Sad) for
same characteristics as the experimental stimuli were used to each of the 935 sounds, considering the total sample (All) and
illustrate the evaluation task and to allow practice making rat- the male and female subsamples separately. We calculated the
ings. After participants had listened to each sound, six indepen- Cronbach’s alpha for all sounds using ratings from 22 partic-
dent emotional rating scales (the SAM and the basic-emotion ipants (α = .95 for valence, α = .92 for arousal, α = .81 for
rating scales) were provided for ratings. The practice session dominance, α = .92 for fear, α = .92 for happiness, and α = .82
lasted approximately 3 min and was conducted to ensure that for sadness) to validate the internal consistency as the previous
the participants were fully familiar with the experiment task. studies did (i.e., Drače, Efendić, Kusturica, & Landzo, 2013;
During the experimental session, each participant was random- Dufey, Fernández, & Mayol, 2011; Suk & Irtel, 2010). The
ly assigned to one of the nine sound groups. The experimental results suggested that the affective ratings for each sound in
task was the same as in the practice session. Each trial was our study were reliable (Iacobucci & Duhachek, 2003). The

Fig. 3 Numbers of sound stimuli in the two-dimensional affective space defined by mean valence and arousal, for (a) IADS-E and (b) IADS-2
1420 Behav Res (2018) 50:1415–1429

Fig. 4 Distributions of overall mean values for the IADS-E sounds in the valence–arousal affective dimensions (a), valence–dominance affective
dimensions (b), and arousal–dominance affective dimensions (c)

coefficient of variation (CV) was also calculated, to evaluate SDmVal = 1.46, SDmDom = 2.05) were obviously lower than in
the variability in the affective ratings; the average values of the the previous research (SDm Aro = 1.92, SDm Val = 1.76,
CVs for the present study across different affective dimen- SDmDom = 1.91). Thus, even though the number of the partic-
sions were CVVal = 31.23%, CVAro = 21.04%, CVDom = ipants was smaller than in the original IADS-2 study, the rat-
19.95%, which revealed a common pattern with the IADS-2 ings in our study were relatively stable.
(CVVal = 36.60%, CVAro = 19.74%, CVDom = 24.66%). We
also compared the standard deviations of the ratings for the Affective Ratings for the IADS-E
IADS-2 sounds between this study and the original one
(Bradley & Lang, 2007a). The mean standard deviations (1) Correlation analyses of valence and arousal We calcu-
(SDms) of the IADS-2 sounds in our study (SDmAro = 1.58, lated the number of sounds in arousal–valence
Behav Res (2018) 50:1415–1429 1421

of the positive sounds (i.e., sounds with valence ratings above

5 points, the midpoint of the 9-point scale used, which are
located in the upper half of Fig. 4a) were distributed evenly
along the arousal dimension. For negative sounds (i.e., sounds
with valence ratings below 5, which are located in the lower
half of the chart), the ratings were more concentrated in the
lower right quadrant of the chart. We divided the sounds into
positive and negative sounds and conducted a correlation
analysis for each. The association between valence and arous-
al was stronger for negative sounds (r = − .51, p < .001) than
for positive sounds, which did not reach the level of statistical
significance (r = − .03, p = .66). This indicates that valence
was independent of arousal for positive sounds, but there was
a strong negative relationship between valence and arousal for
negative sounds. For example, sounds 0379 (Transport), 275
(Scream from IADS-2), and 285 (Attack2 from IADS-2) were
given low valence scores (MVal = 1.28, 1.38, and 1.48, respec-
Fig. 5 A 3-D scatterplot shows the distribution of overall mean values for
the IADS-E in the basic-emotion rating scales tively) and high arousal scores (MAro = 7.80, 7.88, and 7.04,
respectively). The same was not observed for positive sounds:
Sounds 1073 (Music), 1074_b (Music), and 1085 (Music) had
coordinates, with the origin at ratings of 5. The numbers the highest valence scores (MVal = 7.95, 8.05, and 8.09, re-
of sounds in the IADS-E from the first to the fourth spectively), but their arousal scores were not equivalent (MAro
quadrants were 156, 155, 138, and 486, respectively. = 6.77, 3.45, and 7.77, respectively).
Figure 3 more concretely indicates the numbers of
sounds in the two-dimensional affective space defined (2) Correlation analyses of dominance–valence and domi-
by the mean valence and arousal. As compared to the nance–arousal The distributions of the IADS-E ratings
original version of the IADS-2, the numbers of sounds in the valence–dominance and arousal–dominance di-
in all four quadrants are increased. This may make it mensions are showed in Fig. 4b and c, respectively.
easier to choose sounds from each emotional condition. The correlation coefficient analysis showed a strong pos-
Figure 4a illustrates the distributions of the IADS-E stim- itive relationship between valence and dominance (r =
uli in the two-dimensional affective space defined by the .74, R2 = .54, p < .001) and a negative relationship be-
mean valence and arousal of each sound. The valence– tween arousal and dominance (r = − .51, R2 = .26, p <
arousal distribution of the IADS-E sounds was .001). These results indicate that people feel more pow-
boomerang-shaped (i.e., sounds rated as either high pos- erful when they hear positive sounds and feel feeble
itive or high negative tended to be rated with high arousal when they hear high-arousal sounds. Similar trends also
scores), like that obtained in the IADS-2 (Bradley & occurred in the IADS-2 (r = .94, R2 = .88, p < .001, for
Lang, 2007a, Fig. 1). valence and dominance; r = − .54, R2 = .29, p < .001, for
arousal and dominance; see Bradley & Lang, 2007a).
(3) Ratings for basic emotion rating scales We have cate-
The correlation coefficient analysis showed that the linear gorized each sound into three groups by the elicited emo-
correlation between valence and arousal was r = − .40 (R2 = tions’ distribution variabilities based on 90% confidence
.16, p < .001), and the quadratic correlation between the two intervals (CIs), just as in a previous study (Mikels et al.,
dimensions was r = .44 (R2 = .19, p < .001). As compared to 2005). For a given sound, if the mean for one basic emo-
the linear correlation, the contribution ratio of the quadratic tion was higher than the mean of the other two emotions,
correlation was higher and was consistent with those from and if there was no overlap between the CIs for the dom-
previous studies (Choi et al., 2015; Redondo et al., 2008; inant emotion and the remaining emotions, it was
Soares et al., 2013). However, in the quadratic correlation, assigned to a single basic emotion category. If two basic
the correlation value in our study was lower than that found emotions were higher than the mean of the other one
for IADS-2 (r = .47, R2 = .22, p < .001; see Bradley & Lang, basic emotion and if the CIs of the two higher means
2007a). The difference between the two databases could have overlapped with each other, the sound was assigned to
been caused by the difference in their total numbers of stimuli. a blended category. If all three CIs overlapped, the sound
Furthermore, as has been observed in studies using visual was assigned as undifferentiated. As it turned out, 663
stimuli (e.g., Bradley & Lang, 2007b; Lang et al., 2008), most sounds were categorized as single basic emotions—
1422 Behav Res (2018) 50:1415–1429

Table 1 Descriptive statistics, calculated separately for all affective ratings scales in each semantic category

Category Arousal Valence Dominance Fear Happiness Sadness

Animals (N = 54) Mean 5.53 4.76 5.28 3.35 3.66 2.43

SD 0.90 1.42 0.90 1.42 1.41 0.72
Min–Max 2.96–7.17 1.92–7.92 3.13–7.00 1.18–6.29 1.50–6.95 1.27–4.73
People (N = 74) Mean 5.80 3.49 4.82 3.89 2.81 3.17
SD 0.92 1.27 0.99 1.61 1.38 1.47
Min–Max 2.48–7.88 1.18–6.23 2.16–6.77 1.56–7.79 1.09–6.77 1.32–6.86
Nature (N = 70) Mean 4.94 4.70 5.01 3.78 3.04 2.84
SD 1.04 1.14 0.86 1.31 1.04 0.61
Min–Max 3.21–7.83 2.13–7.28 2.77–6.96 1.60–7.27 1.58–5.77 1.67–4.55
Daily Routine Sound (N = 187) Mean 5.51 4.16 5.07 3.74 2.87 2.41
SD 1.10 1.09 1.02 1.30 0.98 0.61
Min–Max 2.86–8.46 1.36–7.08 2.32–7.24 1.46–6.88 1.32–6.05 1.36–4.63
Transport (N = 59) Mean 5.90 3.82 4.38 4.52 2.74 2.65
SD 0.71 0.92 0.76 1.16 0.92 0.79
Min–Max 4.00–7.80 1.28–5.71 2.46–5.82 1.71–7.20 1.20–6.09 1.46–5.28
Electric Sound (N = 64) Mean 6.34 4.05 4.52 4.16 3.58 2.36
SD 0.94 1.06 0.87 1.22 1.26 0.62
Min–Max 4.58–8.33 1.92–6.59 2.46–6.42 1.45–7.04 1.42–6.45 1.32–4.08
Sound Effect (N = 171) Mean 5.23 4.40 4.49 4.48 3.05 3.17
SD 0.97 1.02 0.77 1.09 0.83 0.73
Min–Max 2.96–7.23 1.68–7.05 2.59–6.64 1.80–6.95 1.42–5.76 1.56–5.68
Breaking Sound (N = 56) Mean 6.67 3.24 3.73 5.66 2.18 3.09
SD 0.73 0.57 0.70 0.85 0.40 0.72
Min–Max 4.88–8.08 2.14–5.18 2.50–6.27 2.77–7.36 1.45–3.55 1.41–4.72
Music (N = 170) Mean 5.35 6.30 5.50 2.49 5.04 2.57
SD 1.55 1.07 0.69 1.03 1.38 1.01
Min–Max 1.91–8.21 2.56–8.09 3.09–6.91 1.27–7.14 2.12–8.00 1.09–6.52
Scenarios (N = 30) Mean 6.20 4.67 4.90 3.55 4.13 2.75
SD 0.76 1.39 0.94 1.76 1.79 1.55
Min–Max 4.12–7.33 1.68–7.14 2.00–6.59 1.23–7.92 1.16–6.96 1.20–6.76

All scales range from 1 to 9, in which 1 represents negative, sleepy, being controlled, no fear, no happiness, and no sadness. Min = minimal value; Max =
maximal value; SD = standard deviation; N = number of sounds

happiness (254), fear (391), and sadness (18)—154 as However, unfortunately, the number of sounds that elic-
blended emotion sounds—happiness and fear (87), fear ited sadness was particularly small, and we measured
and sadness (52), and happiness and sadness (15)—and only three basic emotions in this study. These issues need
118 as undifferentiated sounds. For details on the basic to be addressed in future research.
emotion classification for each sound, please refer to the (4) Relationship between the affective dimensions and
supplementary materials. The correlation coefficient basic-emotion rating scales To ascertain the relation-
analysis showed strong negative relationships between ships between the three dimensions (i.e., valence, arous-
fear and happiness (r = − .71, R2 = .51, p < .001) and al, and dominance) and the three basic-emotion rating
sadness and happiness (r = − .49, R2 = .24, p < .001), as scales (i.e., fear, happiness, and sadness), Pearson corre-
well as a positive relationship between fear and sadness lation analyses were performed. The results suggested
(r = .55, R2 = .30, p < .001). The distribution of the that the ratings of valence (rFear = − .82, p < .001; rSad
IADS-E ratings on the three basic-emotion rating scales = − .37, p < .001) and dominance (rFear = − .86, p < .001;
can be seen in Fig. 5 (a 3-D scatterplot). The ratings of rSad = − .43, p < .001) have negative correlations with
basic emotions could assist researchers in choosing emo- fear and sadness, but positive correlations with happi-
tional stimuli more accurately from a discrete categorical ness (rVal = .84, p < .001; rDom = .57, p < .001). The
perspective according to their research purpose. ratings of arousal showed a positive correlation with fear
Behav Res (2018) 50:1415–1429 1423

Table 2 Correlation coefficients resulting from the affective rating scales in each semantic category

Category Dimension Valence Dominance Fear Happiness Sadness

Animals Arousal − .492*** − .442** .433** − .237 − .065

Valence .751*** − .854*** .900*** − .444**
Dominance − .860*** .702*** − .467***
People Arousal − .373** − .619*** .567*** − .092 .446***
Valence .749*** − .612*** .864*** − .610***
Dominance − .819*** .609*** − .695***
Nature Arousal − .693*** − .760*** .749*** − .392** .264*
Valence .746*** − .876*** .862*** − .424***
Dominance − .828*** .554*** − .502***
Daily Routine Sound Arousal − .648*** − .732*** .643*** − .228** .216**
Valence .838*** − .809*** .781*** − .390***
Dominance − .878*** .584*** − .444***
Transport Arousal − .412** − .410** .365** − .128 .202
Valence .711*** − .832*** .839*** − .645***
Dominance − .792*** .585*** − .607***
Electric Sound Arousal − .124 − .400** .125 .181 − .309*
Valence .767*** − .854*** .862*** − .559***
Dominance − .804*** .626*** − .437***
Sound Effect Arousal − .546*** − .562*** .409*** − .081 − .208**
Valence .729*** − .835*** .759*** − .183*
Dominance − .729*** .467*** − .218**
Breaking Sound Arousal − .606*** − .785*** .753*** − .209 .291*
Valence .641*** − .749*** .484*** − .545***
Dominance − .889*** .282* − .487***
Music Arousal − .108 − .237** .073 .604*** − .573***
Valence .629*** − .855*** .612*** − .307***
Dominance − .727*** .351*** − .222**
Scenarios Arousal − .103 − .474** .244 .061 .055
Valence .824*** − .897*** .890*** − .769***
Dominance − .838*** .641*** − .583**
***
p < .001; ** p < .01; * p < .05

(r = .40, p < .001) and a weak negative correlation with rating scales in each semantic category are shown in Table 2.
sadness (r = − .10, p < .01), but no correlation with Although the correlation values are consistent when using all
happiness. As has been demonstrated in previous studies sounds, some subtle differences exist in each semantic catego-
(Choi et al., 2015; Stevenson & James, 2008), valence ry. Considering these results, researchers could choose sounds
and dominance showed positive relationships with hap- from every semantic category in line with their purpose.
piness but negative relationships with the other basic According to these results, we considered that all 935
emotion rating scales (i.e., fear and sadness), whereas sounds (including those from the IADS-2) provide a new ver-
the opposite pattern was observed for arousal. sion of the standardized affective auditory stimulus database
that is capable of effectively evoking emotion. However, the
number of sounds in the low valence–arousal quadrant is par-
Table 1 shows the descriptive statistics for the sound ratings ticularly scarce relative to the other three quadrants, thus sug-
on all affective ratings scales in each semantic category. The gesting that negative stimuli would be rated as much more
ratings in each semantic category have their unique features; for arousing, irrespective of whether they are auditory or visual
example, sounds in music tend to have more positive ratings, stimuli.
whereas breaking sounds tend to have more negative and The ratings of basic emotions could assist researchers in
arousing ratings. The correlation coefficients for the affective choosing emotional stimuli more accurately, according to their
1424 Behav Res (2018) 50:1415–1429

Table 3 Descriptive statistics, calculated separately for each dimension and each semantic category and gender

Male Female

Category Arousal Valence Dominance Arousal Valence Dominance

Animals (N = 54) Mean 5.48 4.87 5.46 5.59 4.64 5.08

SD 1.04 1.52 1.09 0.90 1.40 0.88
Min–Max 2.58–7.36 1.83–8.25 2.17–7.55 3.25–7.25 2.00–8.09 3.42–7.00
People (N = 74) Mean 5.78 3.53 4.91 5.82 3.44 4.71
SD 0.96 1.26 1.10 1.00 1.36 1.01
Min–Max 2.45–7.92 1.09–6.82 2.08–6.73 2.50–7.83 1.18–6.27 2.07–7.00
Nature (N = 70) Mean 4.88 4.83 4.99 5.01 4.56 5.00
SD 1.04 1.11 0.96 1.14 1.26 0.95
Min–Max 2.67–7.83 2.17–7.25 2.83–7.25 3.07–8.09 1.82–7.71 2.64–7.36
Daily Routine Sound (N = 187) Mean 5.46 4.19 5.07 5.56 4.12 5.04
SD 1.11 1.13 1.12 1.18 1.12 1.05
Min–Max 2.18–8.58 1.64–7.09 2.33–7.73 3.25–8.33 1.09–7.57 2.27–7.79
Transport (N = 59) Mean 5.90 4.00 4.38 5.89 3.66 4.37
SD 0.82 0.97 0.89 0.77 0.95 0.79
Min–Max 3.58–8.00 1.27–6.00 2.25–6.64 4.27–7.75 1.29–5.83 2.14–6.07
Electric Sound (N = 64) Mean 6.29 4.20 4.60 6.38 3.88 4.44
SD 1.03 1.15 1.03 0.93 1.07 0.86
Min–Max 4.17–8.42 1.67–7.00 1.83–6.83 4.57–8.25 2.00–7.08 2.57–6.09
Sound Effect (N = 171) Mean 5.19 4.51 4.40 5.27 4.29 4.55
SD 1.03 1.01 0.92 1.02 1.11 0.80
Min–Max 2.58–7.45 1.91–7.27 2.45–6.82 2.64–7.25 1.45–7.21 2.73–6.45
Breaking Sound (N = 56) Mean 6.56 3.42 3.80 6.77 3.07 3.66
SD 0.76 0.63 0.86 0.81 0.64 0.75
Min–Max 4.67–8.25 2.36–5.27 2.55–6.36 5.08–7.93 1.73–5.09 2.36–6.18
Music (N = 170) Mean 5.28 6.36 5.45 5.42 6.22 5.54
SD 1.67 1.04 0.83 1.50 1.17 0.74
Min–Max 1.92–8.50 2.82–8.08 2.00–7.36 1.75–8.17 2.21–8.33 2.93–7.17
Scenarios (N = 30) Mean 6.30 4.79 5.01 6.11 4.55 4.80
SD 0.84 1.34 0.91 0.82 1.50 1.12
Min–Max 3.73–7.92 1.91–7.18 2.64–6.64 4.43–7.58 1.50–7.09 1.50–6.79

All scales range from 1 to 9, in which 1 represents negative, sleepy, being controlled, no fear, no happiness, and no sadness. Min = minimal value; Max =
maximal value; SD = standard deviation; N = number of sounds

research purpose. Unfortunately, because of the long experi- = .00, did not reach statistical significance here, but the ratings
mental time, we measured only three basic emotions in this of valence, F(1, 1868) = 8.27, p = .004, η2 = .004; happiness,
study. Further research should measure other basic emotions, F(1, 1868) = 29.49, p < .001, η2 = .02; and sadness, F(1, 1868)
such as disgust, anger, and surprise. = 31.47, p < .001, η2 = .02, did show significant differences
between the genders. Table 3 shows the descriptive statistics
Sex difference in sounds ratings for valence, arousal, and dominance for the female and male
subsamples in each semantic category separately. The results
A multivariate analysis of variance (MANOVA) was conduct- suggest that Japanese men assessed the sounds as being much
ed with between-subjects factors to determine whether differ- more emotional than did the women; however, the difference
ences existed between the genders (male and female) for all according to gender was particularly small.
affective rating scales with all sounds. The results suggested Pearson’s correlations were calculated in order to ex-
that the main effects of gender in the ratings of arousal, F(1, amine the relationship between the ratings of arousal, va-
1868) = 2.61, p = .11, η2 = .001; dominance, F(1, 1868) = lence, and dominance for each semantic category and gen-
0.28, p = .60, η2 = .00; and fear, F(1, 1868) = 0.23, p = .63, η2 der separately (Table 4). The relationship between every
Behav Res (2018) 50:1415–1429 1425

Table 4 Correlation coefficients resulting from correlations of ratings of valence, arousal, and dominance, listed for each category and gender

Male Female

Category Dimension Valence Arousal Valence Arousal

Animals Arousal − .597*** − .297*

Dominance .705*** − .421** .712*** − .347*
People Arousal − .254* − .392**
Dominance .699*** − .483*** .707*** − .559***
Nature Arousal − .592*** − .725***
Dominance .630*** − .551*** .761*** − .779***
Daily Routine Sound Arousal − .533*** − .666***
Dominance .746*** − .584*** .806*** − .739***
Transport Arousal − .236 − .510***
Dominance .623*** − .139 .687*** − .513***
Electric Sound Arousal − .028 − .215
Dominance .710*** − .313* .705*** − .468***
Sound Effect Arousal − .451*** − .545***
Dominance .608*** − .398*** .668*** − .529***
Breaking Sound Arousal − .511 ***
− .587***

Dominance .598*** − .649*** .505*** − .744***

Music Arousal − .080 − .112
Dominance .534*** − .120 .613*** − .268***
Scenarios Arousal .155 − .273
Dominance .638*** − .336 .835*** − .475**
***
p < .001; ** p < .01; * p < .05

Table 5 List of sounds that differ by more than 2.5 standard deviations between male and female participants’ ratings in the valence and arousal
dimensions for the IADS-E

Sound ID Male Female Z-Distance

Source Description Arousal Valence Arousal Valence

Expanded HorrorFilm23 0082 6.18 4.73 6.55 2.82 2.51

Expanded Pops14 1098 7.50 6.08 6.17 4.67 2.52
IADS-2 RattleSnake 134 5.73 5.73 4.55 4.18 2.52
Expanded Iron Sheet 0876 6.45 5.18 6.00 3.27 2.55
Expanded SuspenseFilm14 0068 4.00 6.67 5.58 5.42 2.68
Expanded Cellophane Tape 0639 4.91 3.18 6.91 2.91 2.68
Expanded Bell1 0263 5.00 6.42 7.00 6.75 2.70
Expanded Birds4 0420 4.73 5.27 3.71 7.07 2.78
Expanded HorrorFilm21 0058 6.64 4.73 6.64 2.64 2.84
Expanded HorrorFilm17 1374 5.64 5.36 6.55 3.45 2.90
Expanded VideoGame7 0768 6.27 3.73 7.91 2.36 2.93
Expanded SuspenseFilm15 0076 3.75 6.17 5.58 4.67 3.46
IADS-2 SlotMachine1 716 7.45 7.00 5.82 5.18 3.64
Expanded Birds1 0346 3.50 7.83 5.83 6.92 3.77
Expanded Sea Wave2 0337 6.00 4.27 7.55 2.00 4.31
Expanded Sea1 0123 3.25 7.25 6.08 5.75 5.34

Z-Distance is the distance between male and female participants’ emotional ratings on the valence and arousal dimensions, which is converted into a z-
score
1426 Behav Res (2018) 50:1415–1429

Table 6 Proportions of sound stimuli in each valence condition, defined by loudness of the peak sound intensity

IADS-E IADS-2

Valence Category Physical Intensity Low Moderate High Low Moderate High
(≤70 dB) (70–75 dB) (≥75) (≤70 dB) (70–75 dB) (≥75)

Positive 8.1% 5.8% 2.8% 3.6% 4.8% 6.6%

Neutral 21.0% 14.2% 9.0% 13.2% 16.2% 12.0%
Negative 7.1% 14.1% 17.5% 5.4% 15.0% 23.4%

IADS-E is the sum for both the new sounds and the IADS-2 sounds (N = 935)

two affective dimensions showed differences in each se- − .31 (p < .001) for valence, r = .47 (p < .001) for arousal, r = −
mantic category between males and females. We also .39 (p < .001) for dominance, r = .32 (p < .001) for fear, r = −
identified differences in the affective ratings for each .10 (p = .003) for happiness, and r = .18 (p < .001) for sadness.
sound between the genders. The distance between male The relationship between valence and the peak intensity
and female participants’ affective ratings in the dimen- across the 935 sounds was inconsistent with Bradley and
sions of valence and arousal was measured for each of Lang’s (2000) research, which suggested that there is no cor-
the 935 sounds and converted into a z-score. A large z- relation between valence and peak sound intensity. On the
score indicates a greater difference in ratings between the other hand, arousal had a significant positive relationship with
male and female participants. Table 5 lists 16 sounds for peak sound intensity, and this result is consistent with Bradley
which the standardized distance was greater than 2.5 stan- and Lang’s (2000) study. To further clarify the relationship
dard deviations. between the affective ratings and physical intensity, quadratic
As we mentioned above, even though no pronounced correlations were calculated between peak sound intensity and
differences were found in gender, there are still some spe- affective ratings. The results showed that only valence (r =
cific individual sounds’ ratings showed big gaps in gen- .34, p < .001) and happiness (r = .17, p < .001) reached the
der. We should pay attention to these sounds when level of significance; however, these correlations accounted
selecting the experimental materials. for only 11.5% and 3%, respectively, of the valence and hap-
piness variance. These results revealed that physically intense
Physical intensity sounds were more likely to be highly arousing and to cause a
fear response, but the relationship between a sound’s physical
Correlation analyses were performed to identify the relation- intensity and valence looks more complex.
ship between sounds’ physical intensities and their affective To clarify this problem, we first divided the sounds into
ratings. The Pearson correlation coefficients between a three conditions according to the valence rating (ratings higher
sound’s peak intensity (Lmax) and affective ratings were r = than 6 for positive sounds, greater than or equal to 4 and less
than or equal to 6 for neutral sounds, and less than 4 for
negative sounds). Then we calculated the percentages of the
Table 7 Descriptive statistics, calculated separately for each affective
sounds in three different physical intensity ranges: low inten-
rating scale in our study and the original version, for all IADS-2 sounds
sity (peak intensity ≤ 70 dB), moderate intensity (70 dB <
Dimension Our Study Original Version peak intensity < 75 dB), and high intensity (peak intensity ≥
75 dB) under the three valence types (Table 6). For sounds
Valence Mean 4.40 4.78
categorized as having positive or neutral valence, there were
SD 1.54 1.76
more sounds with low and moderate than with high intensities,
Min–Max 1.18–7.86 1.57–7.90
whereas for negative valence, there were more sounds with
Arousal Mean 5.85 5.84 high and moderate than with low intensities. This clarifies that
SD 1.00 1.16 the positive sounds are usually mild, whereas the negative
Min–Max 2.48–8.00 2.88–8.16 sounds are usually perceived as being intense and strong.
Dominance Mean 5.04 4.71 However, the numbers of sounds in the IADS-2 for each in-
SD 1.04 1.17 tensity were distributed relatively uniformly. On the basis of
Min–Max 2.00–7.24 2.29–6.86 the results in Table 6, we suspect that when people make an
All scales range from 1 to 9, in which 1 represents negative, sleepy, and
evaluation for the valence dimension, Americans may focus
being controlled. Min = minimal value; Max = maximal value; SD = on both the physical intensity of sounds and the context in
standard deviation which the sound is expressed; Japanese participants, on the
Behav Res (2018) 50:1415–1429 1427

Table 8 List of sounds that differ by more than two standard deviation between this study and the original version in the valence and arousal
dimensions for the IADS-2

Description Sound ID Japan Original Z-Distance

Arousal Valence Arousal Valence

BoyLaugh 220 4.56 6.33 7.28 6.00 2.13

Thunderstorm 602 3.61 5.17 5.99 3.77 2.14
Party 365 4.33 5.50 6.97 6.32 2.15
RollerCoaster 360 4.17 7.06 6.94 7.54 2.22
EroticFem2 202 4.11 6.28 6.81 7.13 2.24
EroticCouple3 216 3.00 6.17 5.97 6.84 2.53
Belch 702 1.44 6.06 4.45 5.37 2.58
EroticFem1 201 3.94 5.89 6.7 7.31 2.60
EroticCouple2 215 3.50 6.11 6.47 7.32 2.74
EroticMale1 210 2.61 4.39 5.72 6.64 3.59

Z-Distance is the distance between our study and the original version’s emotional ratings on the valence and arousal dimensions, which is converted into
a z-score

other hand, may pay more attention to the content of the (220), a party (365), a roller coaster (360), and a belch (702),
sounds. This interpretation should be considered in other sub- showed no difference in the arousal dimension; however, for
sequent research. valence, the original raters rated them as more positive than in
this study. For the sound of a thunderstorm (602), our partic-
Comparison of affective ratings for the IADS-2 sounds ipants rated it as more negative and higher in arousal than did
those in the original version.
We compared the affective ratings of the IADS-2 sounds in the Overall, although we must note carefully that our partici-
original version (Bradley & Lang, 2007a) with the results in pants evaluated the IADS-2 sounds in a highly similar way to
our study. Table 7 presents the means, standard deviations, the the participants in the original version, it is also true that some
minimal and maximal values for the ratings of arousal, va- slight differences have been found. These differences should
lence, and dominance from our study and the original study not be overlooked, particularly for the sounds closely related
separately. Very high Pearson correlations were obtained for to human activities.
all three ratings, of valence, arousal, and dominance, between
the two groups (rVal = .81, p < .001; rAro = .76, p < .001; rDom
= .71, p < .001). The results indicate that the two groups of Conclusion
participants rated the IADS-2 sounds in similar manners.
To evaluate the differences in each affective dimension, a In this article, we collected 935 affective sounds based on the
MANOVA was conducted with culture as a between-subjects IADS-2 and tried to expand and improve the existing affective
factor and with the affective dimensions as dependent vari- auditory stimuli database. The results showed that we could
ables. The ratings of arousal, F(1, 332) = 0.002, p = .96, η2 = successfully provide a larger standardized database of
.00, did not reach the level of statistical significance here, but emotional-evocative stimuli, which includes contents across
the ratings of dominance, F(1, 332) = 7.44, p < .01, η2 = .02, a wider range of semantic categories. Through our research,
and valence, F(1, 332) = 4.58, p < .05, η2 = .01, showed the new affective sound stimulus standardized database
significant differences between the two groups. The (IADS-E) may enable the comparison of results across various
Japanese participants rated sounds as less positive and more kinds of auditory research.
dominant than did the original participants (Table 7). To clar- On the other hand, it is obvious that real-life emotional
ify in more detail the differences for individual sounds in the experiences mostly rely on the presence of combined stimuli
IADS-2, we calculated the distances in the dimensions of va- coming from multiple sensory modalities (Baumgartner,
lence and arousal for each of the 167 sounds between our Esslen, & Jäncke, 2006). Information received may have been
study and the original version, and then converted each dis- transmitted through different modalities and channels.
tance into a z-score. Table 8 lists ten sounds for which the Auditory stimuli can increase the power of affective visual
standardized distance was greater than two standard devia- stimuli, and vice versa. The visual and auditory modalities
tions. The sounds closely related to human activities, such as can integrate to provide a complete assessment of the emo-
erotic sounds (202, 216, 201, 215, and 210), a boy laughing tional qualities of a situation or an object (Gerdes et al., 2014).
1428 Behav Res (2018) 50:1415–1429

Through this article, we expected to be able to provide a reli- research on auditory affective processing. Behavior Research
Methods, 40, 531–539. https://doi.org/10.3758/BRM.40.2.531
able auditory stimulus database to examine the mechanism of
Bradley, M., & Lang, P. J. (1994). Measuring emotion: The Self-
emotional processing of different modalities and promote this Assessment Manikin and the semantic differential. Journal of
kind of emotional research. Behavior Therapy and Experimental Psychiatry, 25, 49–59.
As we mentioned earlier, there are still some limitations to https://doi.org/10.1016/0005-7916(94)90063-9
Bradley, M. M., & Lang, P. P. J. (1999). International Affective Digitized
our study. First, as compared with a visual stimulus database
Sounds (IADS-1): Stimuli, instruction manual, and affective ratings
such as the IAPS (Lang et al., 2008), although the IAPS did (Technical Report No. B-2). Gainesville, FL: University of Florida,
not make the number of stimuli in each arousal–valence quad- Center for Research in Psychophysiology.
rant equal either, it is obvious that the IAPS’s distribution is Bradley, M. M., & Lang, P. J. (2000). Affective reactions to acoustic
stimuli. Psychophysiology, 37, 204–215. https://doi.org/10.1111/
relatively balanced. In response to this study, we are commit-
1469-8986.3720204
ted to increasing the number of sounds for the quadrants that Bradley, M. M., & Lang, P. J. (2007a). The International Affective
have few sounds and trying to develop a more practical affec- Digitized Sounds: Affective ratings of sounds and instruction
tive sound stimulus database. Second, though the auditory manual (Technical Report No. B-3). Gainesville, FL: University of
Florida, NIMH Center for the Study of Emotion and Attention.
stimuli were less sensitive than visual stimuli are to gender
Bradley, M. M., & Lang, P. J. (2007b). Affective Norms for English Words
influences, there were still some individual sound stimuli with (ANEW): Instruction manual and affective ratings (Technical
large differences in gender. Therefore, when we select exper- Report No. C-1). Gainesville, FL: University of Florida, Center for
imental stimuli for psychological experiments, it is necessary Research in Psychophysiology.
to pay attention to the specific individual sounds. Third, one Buodo, G., Sarlo, M., Mento, G., Messerotti Benvenuti, S., & Palomba,
D. (2017). Unpleasant stimuli differentially modulate inhibitory pro-
might want to argue that the new database is just a local data- cesses in an emotional Go/NoGo task: An event-related potential
base for Japanese research, because our main participants study. Cognition and Emotion, 31, 127–138. https://doi.org/10.
were Japanese college students. Even though we have deter- 1080/02699931.2015.1089842
mined that the participants in our study and in the original Castonguay, A. L., Sabiston, C. M., Crocker, P. R. E., & Mack, D. E.
(2014). Development and validation of the Body and Appearance
version rated sound stimuli in highly similar manners, whether Self-Conscious Emotions Scale (BASES). Body Image, 11, 126–
or not the emotional ratings obtained in this study would be 136. https://doi.org/10.1016/j.bodyim.2013.12.006
applicable to the elderly, children, or those in other countries Castro, S. L., & Lima, C. F. (2010). Recognizing emotions in spoken
remains to be explored in further studies. Fourth, in this study language: A validated set of Portuguese sentences and
pseudosentences for research on emotional prosody. Behavior
we assessed only three basic emotions (happiness, sadness, Research Methods, 42, 74–81. https://doi.org/10.3758/BRM.42.1.
and fear); to offer a more complete characterization of the 74
basic emotions aroused by the IADS-E stimuli, further re- Choi, Y., Lee, S., Choi, I. M., Jung, S., Park, Y. K., & Kim, C. (2015).
search should measure other basic emotions, such as disgust, International Affective Digitized Sounds in Korea: A cross-cultural
adaptation and validation study. Acta Acustica United with Acustica,
anger, and surprise. 101, 134–144. https://doi.org/10.3813/AAA.918811
Choi, Y., Lee, S., Jung, S., Choi, I.-M., Park, Y.-K., & Kim, C. (2016).
Author note This research was supported by the Center of Innovation Erratum to: Development of an auditory emotion recognition func-
Program of the Japan Science and Technology Agency (JST). In addition, tion using psychoacoustic parameters based on the International
we thank Syouichi Shiota, Madoca Miyagi, and Shiho Kashihara for their Affective Digitized Sounds. Behavior Research Methods, 48, 827.
kind advice on the manuscript and analysis. Special thanks to Yuko https://doi.org/10.3758/s13428-015-0596-x
Sonobe, Shogo Aida, Narumi Yamagata, Masashi Nishijima, Aiko Czigler, I., Cox, T. J., Gyimesi, K., & Horváth, J. (2007). Event-related
Nagao, and Noriko Miura for their assistance with the experiments and potential study to aversive auditory stimuli. Neuroscience Letters,
data collection. 420, 251–256. https://doi.org/10.1016/j.neulet.2007.05.007
da Silva, S. P., & Backs, R. W. (2015). Cardiac response during auditory
selective attention to tones and affective sounds. Psychophysiology,
52, 1099–1105. https://doi.org/10.1111/psyp.12432
References Drače, S., Efendić, E., Kusturica, M., & Landzo, L. (2013). Cross-cultural
validation of the Binternational affective picture system^(IAPS) on a
sample from bosnia and herzegovina. Psihologija, 46, 17–26. doi:
Armony, J. L., Aubé, W., Angulo-Perkins, A., Peretz, I., & Concha, L. https://doi.org/10.2298/PSI1301017D
(2015). The specificity of neural responses to music and their rela- Droit-Volet, S., Ramos, D., Bueno, J. L. O., & Bigand, E. (2013). Music,
tion to voice processing: An fMRI-adaptation study. Neuroscience emotion, and time perception: The influence of subjective emotional
Letters, 593, 35–39. https://doi.org/10.1016/j.neulet.2015.03.011 valence and arousal? Frontiers in Psychology, 4, 417:1–12. doi:
Banse, R., & Scherer, K. R. (1996). Acoustic profiles in vocal emotion https://doi.org/10.3389/fpsyg.2013.00417
expression. Journal of Personality and Social Psychology, 70, 614– Dufey, M., Fernández, A. M., & Mayol, R. (2011). Adding support to
636. https://doi.org/10.1037/0022-3514.70.3.614 cross-cultural emotional assessment: Validation of the International
Baumgartner, T., Esslen, M., & Jäncke, L. (2006). From emotion percep- Affective Picture System in a Chilean sample. Universitas
tion to emotion experience: Emotions evoked by pictures and clas- Psychologica, 10, 521–533.
sical music. International Journal of Psychophysiology, 60, 34–43. Ekman, P. (1992). Are there basic emotions? Psychological Review, 99,
https://doi.org/10.1016/j.ijpsycho.2005.04.007 550–553. https://doi.org/10.1037/0033-295X.99.3.550.
Belin, P., Fillion-Bilodeau, S., & Gosselin, F. (2008). The Montreal Esslen, M., Pascual-Marqui, R. D., Hell, D., Kochi, K., & Lehmann, D.
Affective Voices: A validated set of nonverbal affect bursts for (2004). Brain areas and time course of emotional processing.
Behav Res (2018) 50:1415–1429 1429

NeuroImage, 21, 1189–1203. https://doi.org/10.1016/j.neuroimage. Marchewka, A., Zurawski, L., Jednoróg, K., & Grabowska, A. (2014).
2003.10.001 The Nencki Affective Picture System (NAPS): Introduction to a
Fabiani, M., Kazmerski, V. A., Cycowicz, Y. M., & Friedman, D. (1996). novel, standardized, wide-range, high-quality, realistic picture data-
Naming norms for brief environmental sounds: Effects of age and base. Behavior Research Methods, 46, 596–610. https://doi.org/10.
dementia. Psychophysiology, 33, 462–475. 3758/s13428-013-0379-1
Fazio, R. H. (2001). On the automatic activation of associated evalua- Marin, M. M., Gingras, B., & Bhattacharya, J. (2012). Crossmodal trans-
tions: An overview. Cognition and Emotion, 15, 115–141. https:// fer of arousal, but not pleasantness, from the musical to the visual
doi.org/10.1080/0269993004200024 domain. Emotion, 12, 618–631. https://doi.org/10.1037/a0025020
Gerdes, A. B. M., Wieser, M. J., & Alpers, G. W. (2014). Emotional Mikels, J. A., Fredrickson, B. L., Larkin, G. R., Lindberg, C. M., Maglio,
pictures and sounds: A review of multimodal interactions of emotion S. J., & Reuter-Lorenz, P. A. (2005). Emotional category data on
cues in multiple domains. Frontiers in Psychology, 5, 1351:1–10. images from the International Affective Picture System. Behavior
https://doi.org/10.3389/fpsyg.2014.01351 Research Methods, 37, 626–630. https://doi.org/10.3758/
Iacobucci, D., & Duhachek, A. (2003). Advancing alpha: Measuring BF03192732
reliability with confidence. Journal of Consumer Psychology, 13, Oldfield, R. C. (1971). The assessment and analysis of handedness: The
478–487. https://doi.org/10.1207/S15327663JCP1304_14 Edinburgh inventory. Neuropsychologia, 9, 97–113. https://doi.org/
Iwata, N., & Higuchi, H. R. (2000). Responses of Japanese and American 10.1016/0028-3932(71)90067-4
university students to the STAI items that assess the presence or Pell, M. D., & Kotz, S. A. (2011). On the time course of vocal emotion
absence of anxiety. Journal of Personality Assessment, 74, 48–62. recognition. PLoS ONE, 6, e27256. https://doi.org/10.1371/journal.
https://doi.org/10.1207/S15327752JPA740104 pone.0027256
Jaquet, L., Danuser, B., & Gomez, P. (2012). Music and felt emotions: Redondo, J., Fraga, I., Padrón, I., & Piñeiro, A. (2008). Affective ratings
How systematic pitch level variations affect the experience of pleas- of sound stimuli. Behavior Research Methods, 40, 784–790. https://
antness and arousal. Psychology of Music, 42, 51–70. https://doi. doi.org/10.3758/BRM.40.3.784
org/10.1177/0305735612456583 Schindler, S., & Kissler, J. (2016). Selective visual attention to emotional
Juslin, P. N., & Laukka, P. (2003). Communication of emotions in vocal words: Early parallel frontal and visual activations followed by in-
expression and music performance: Different channels, same code? teractive effects in visual cortex. Human Brain Mapping, 37, 3575–
Psychological Bulletin, 129, 770–814. https://doi.org/10.1037/ 3587. https://doi.org/10.1002/hbm.23261
0033-2909.129.5.770 Schirmer, A., & Kotz, S. A. (2006). Beyond the right hemisphere: Brain
Krosnick, J. A., & Presser, S. (2010). Question and questionnaire design. mechanisms mediating vocal emotional processing. Trends in
In P. V. Marsden & J. D. Wright (Eds.), Handbook of survey re- Cognitive Sciences, 10, 24–30. https://doi.org/10.1016/j.tics.2005.
search (pp. 263–314). Bingley, UK: Emerald Group. 11.009
Lang, P. J. (1980). Behavioral treatment and bio-behavioral assessment: Shimizu, H., & Imae, K. (1981). Development of Japanese collegiate
Computer applications. In J. B. Sidowski, J. H. Johnson, & T. A. version of State–Trait Anxiety Inventory (in Japanese). Japanese
Williams (Eds.), Technology in mental health care delivery (pp. Journal of Educational Psychology, 29, 62–67.
119–137). Norwood, NJ: Ablex. Shuman, V., Clark-Polner, E., Meuleman, B., Sander, D., & Scherer, K.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1997). International R. (2015). Emotion perception from a componential perspective.
Affective Picture System (IAPS): Technical manual and affective Cognition and Emotion, 31, 47–56. https://doi.org/10.1080/
ratings (Technical Report No. A-1). Gainesville, FL: University of 02699931.2015.1075964
Florida, NIMH Center for the Study of Emotion and Attention. Soares, A. P., Pinheiro, A. P., Costa, A., Frade, C. S., Comesaña, M., &
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2008). International Pureza, R. (2013). Affective auditory stimuli: Adaptation of the
affective picture system (IAPS): Affective ratings of pictures and International Affective Digitized Sounds (IADS-2) for European
instruction manual (Technical Report A-8). Gainesville, FL: Portuguese. Behavior Research Methods, 45, 1168–1181. https://
University of Florida, Center for Research in Psychophysiology. doi.org/10.3758/s13428-012-0310-1
Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (1993). Spielberger, C.D., Gorsuch, R. L., & Lushene, R. E. (1970). STAI:
Looking at pictures: Affective, facial, visceral, and behavioral reac- Manual for the State–Trait Anxiety Inventory (BSelf-Evaluation
tions. Psychophysiology, 30, 261–273. https://doi.org/10.1111/j. Questionnaire^). Palo Alto, CA: Consulting Psychologists Press.
1469-8986.1993.tb03352.x Stevenson, R. A, & James, T. W. (2008). Affective auditory stimuli:
Lindquist, K. A. (2010). The brain basis of emotion: A meta-analytic Characterization of the International Affective Digitized Sounds
review. Dissertation Abstracts International, B: Sciences and (IADS) by discrete emotional categories. Behavior Research
Engineering, 71, 2744. https://doi.org/10.1017/ Methods, 40, 315–321. https://doi.org/10.3758/BRM.40.1.315
S0140525X11000446 Suk, H. J., & Irtel, H. (2010). Emotional response to color across media.
Liu, P., & Pell, M. D. (2012). Recognizing vocal emotions in Mandarin Color Research and Application, 35, 64–77. https://doi.org/10.
Chinese: A validated database of Chinese vocal emotional stimuli. 1002/col.20554
Behavior Research Methods, 44, 1042–1051. https://doi.org/10. Vieillard, S., Peretz, I., Gosselin, N., Khalfa, S., Gagnon, L., & Bouchard,
3758/s13428-012-0203-3 B. (2008). Happy, sad, scary and peaceful musical excerpts for re-
Lolli, S. L., Lewenstein, A. D., Basurto, J., Winnik, S., & Loui, P. (2015). search on emotions. Cognition and Emotion, 22, 720–752. https://
Sound frequency affects speech emotion perception: Results from doi.org/10.1080/02699930701503567
congenital amusia. Frontiers in Psychology, 6, 1340:1–10. https:// Zillmann, D. (1988). Mood management through communication
doi.org/10.3389/fpsyg.2015.01340 choices. American Behavioral Scientist, 31, 327–340.