Assessing Emotion-Induced Variations of Event-Related Potentials

and Heart Rate During Affective Picture Processing

Stefania Coelli

, Pierluigi Reali

and Anna Maria Bianchi

Department of Electronics Information and Bioengineering, Politecnico di Milano, Milano, Italy

Keywords: Event-Related Potential, Heart Rate, Emotions.

Abstract: Emotions are psychological responses to stimuli that can induce measurable variations in physiological

parameters. While actual emotions span a continuum spectrum, they can be grouped into a finite number of

classes or modeled in terms of independent dimensions, the most common of which are arousal (low to high)

and valence (positive, neutral, and negative). In this work, we investigated the modulation of physiological

parameters related to both the central (CNS) and the autonomic (ANS) nervous systems induced by passive

and sustained affective stimulation. Specifically, an Event-Related Potential (ERP) analysis was conducted to

explore the effect of the arousal and valence dimensions on cortical activation. Meanwhile, their influence on

the ANS activity was evaluated through time-domain heart rate (HR) parameters. When high arousal stimuli

are delivered, the experiment revealed that specific ERP components (i.e., P300 and the late positive potential,

LPP) are modulated by the valence dimension, with positive and negative images inducing a stronger response

than neutral stimuli. Instead, the early posterior negativity (EPN) was found to be influenced by the stimulus

arousal but not by the valence of the processed pictures. Finally, HR parameters were principally modulated

by the valence of the stimulation, in line with the observed ERP changes and expectations from the literature.

1 INTRODUCTION

Emotions can be described as the responses to

external or internal stimuli influenced by individual

experiences that are able to induce physiological

changes (Reali et al., 2018b). Both the central nervous

system (CNS) and the autonomic nervous system

(ANS) have a fundamental role in the ability to

regulate emotions during the processing of affective

stimuli (Mizuno-Matsumoto et al., 2020). Despite the

great interest in human emotions, few conclusions

about the emotional response of these two systems

had been drawn until a standardization of the

stimulation was proposed. Based on the first affective

and psychological research theories and findings, a

dimensional model of emotional states has been

derived and widely employed. According to this

model, all the affective states can be seen as the linear

combination of valence (pleasant/positive vs

unpleasant/negative sensation) and arousal

(calm/low vs excited/high) components (Russell,

2003). Following this perspective, the International

https://orcid.org/0000-0002-9607-8755

https://orcid.org/0000-0003-3041-4004

https://orcid.org/0000-0002-8290-7460

Affective Picture System (IAPS) dataset has been

developed, validated, and continuously updated to

provide a standardized set of emotional stimuli for

affective research (Lang Bradley, M.M., & Cuthbert,

B.N., 2008).

The study of brain activity modulations in

response to emotional stimuli has a long history in the

literature, particularly in the form of event-related

potential (ERP) analysis (Olofsson et al., 2008), and

has often been paired, or substituted, with the

evaluation of changes in parameters related to the

ANS (Polo et al., 2023; Reali et al., 2018a; Telles et

al., 2019).

In terms of ERP analysis, findings consistently

identified the early posterior negativity (EPN) as

modulated by the arousal level of the stimulus, the

late positive potential (LPP) component as associated

with the processing of the emotional information

(thus related to the valence dimension), while the

P300 peak amplitude appeared modulated by the

combination of valence and arousal dimensions

(Olofsson et al., 2008; Schindler et al., 2022).

Coelli, S., Reali, P. and Bianchi, A.

Assessing Emotion-Induced Variations of Event-Related Potentials and Heart Rate During Affective Picture Processing.

DOI: 10.5220/0012430400003657

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2024) - Volume 1, pages 667-674

ISBN: 978-989-758-688-0; ISSN: 2184-4305

667

Regarding a more general assessment of cortical

response, a prominent frontal activation, also in terms

of an asymmetry in the alpha frequency band, has

often been reported in response to positive and

negative stimulation (Coan and Allen, 2004; Kop et

al., 2011).

However, due to differences in the study

protocols (i.e., passive/active viewing, exposure

times, pictures or video stimuli, etc.), the objectives,

the devices used to deliver the stimuli (e.g., PC

monitors, virtual reality visors, immersive settings) or

to collect the physiological responses, and, above all,

the variability between subjects, results are difficult

to replicate across studies.

Heart rate variability (HRV) has often been used

as an indicator of the ANS response to emotions, both

in the time and frequency domains. Together with

other ANS-related parameters, such as respiration

rate and electrodermal activity, the heart rate (HR)

variations have been used to define emotion

classification models (Egger et al., 2019; Polo et al.,

2023; Reali et al., 2018a). Moreover, HR parameters

in the time domain (i.e., mean and standard deviation)

at rest have been explored to determine the emotional

regulation ability of the subjects and predict their

cortical response to successive emotional

stimulations or attentional tasks (Ruiz-Padial and

Mercado, 2021; Telles et al., 2019), but rarely they

have been analyzed in conjunction with ERPs

variations. Indeed, to the best of our knowledge, no

studies explored the influence of both valence and

arousal factors on ERP components and HR

simultaneously.

In this work, we present an ERP analysis and an

evaluation of the simultaneous ANS response during

affective picture stimulation. Specifically, we

designed an IAPS-based passive stimulation protocol

featuring sequences of images delivered for an

extended exposure time (12 seconds). The goal of

such a long exposure, unusual for an ERP analysis,

was to allow the volunteer to process each affective

stimulus completely and test the effect of this

processing rather than the first impression of each

picture. Moreover, this was required to properly

evaluate ANS variations, as analysis windows of

adequate length are needed to accurately estimate HR

parameters (Castaldo et al., 2019). As for the ERP

analysis, we focused on the late components (EPN,

P300, and LPP), which have been associated with the

encoding processing phase for emotional stimuli

(Olofsson et al., 2008).

2 MATERIALS AND METHODS

2.1 Participants and Procedure

Thirty-one healthy participants (22 males and 9

females) aged between 19 and 27 (mean=20.9;

SD=1.5) were recruited for the study. Volunteers

were instructed about the protocol and told they could

withdraw from the trial at any moment. Before the

beginning of each experimental session, participants

were asked to sign a written consent to participate in

the study.

The experiment was conducted in front of a PC

monitor, which was used to provide the visual stimuli.

Ninety images were selected from the IAPS database

and grouped into three levels of arousal and three of

valence. Specifically, the selected arousal ranges

were 2.41-3.59 for the “low arousal” (LA) set of

pictures, 3.60-5.49 for the “medium arousal” (MA)

one, and 5.50-7.35 for the “high arousal” (HA) set.

Within each arousal block, three subsets of valence

pictures, namely “negative,” “neutral,” and “positive”

valence, were selected, containing ten pictures each

and showing the following valence ranges: 2.04-5.39

(negative), 5.40-6.19 (neutral), and 6.20-8.28

(positive). Combining these arousal and valence

ranges led to nine arousal-valence partitions, each

containing ten pictures expected to elicit similar

emotions. The above ranges were chosen as a trade-

off to maximize the separation among the nine sets of

images in terms of arousal and valence while

preserving the desired number of pictures (ten) for

each of the nine arousal-valence level combinations.

Regarding the stimulation protocol (Figure 1), the

nine arousal-valence blocks were presented in order

of increasing levels of arousal to keep participants

engaged during the experiment. To avoid any other

sequence-related bias, the three valence sub-blocks

(i.e., negative, neutral, and positive) and the ten

pictures of each sequence were presented in a random

order. A “neutral image”, a picture intended not to

elicit any particular emotion showing a country

landscape without any particular subject represented,

was shown for 30 seconds between each sequence of

ten pictures to reduce the propagation of the

physiological response from one arousal-valence

block to the following.

Each IAPS image was displayed for 12 seconds;

therefore, each condition had a total duration of 2

minutes. A first 90-second neutral image was

displayed at the beginning of the protocol to make

participants relax and prepare them for the

presentation of the first arousal-valence block.

BIOSIGNALS 2024 - 17th International Conference on Bio-inspired Systems and Signal Processing

668

Figure 1: Stimulation sequence. Arousal blocks of pictures were always presented in increasing order. Within each arousal

block, the valence sub-blocks were presented in randomized order (i.e., different for each participant). The ten pictures

presented within each valence block were also randomized.

Electroencephalographic (EEG) and electrocar-

diographic (ECG) signals were collected during the

entire procedure.

The protocol was approved by the Institutional

Ethics Committee of Politecnico di Milano.

2.2 Signal Acquisition

EEG signal was acquired by means of a portable

system (SD LTM Express and System Plus Evolution

software, Micromed, Italy). Brain signals were

recorded from 25 electrodes placed according to the

10/10 system with the following recording channels:

Fpz, Fp1, Fp2, AF3, AF4, AF7, AF8, Fz, F1, F2, F3,

F4, F5, F6, F7, F8, Cz, C3, C4, T7, T8, P7, P8, O1

and O2. Data were acquired at a sampling rate of 256

Hz, with the reference electrode placed on the pre-

cabled cap, between CPz and Pz.

The ProComp Infiniti system (Thought

Technology Ltd., Canada), a battery-powered device,

was used to record a single lead ECG at the default

sampling rate of 2048 Hz, with three pre-gelled

disposable electrodes placed in lead II configuration.

2.3 ECG and EEG Signal Analysis

The collected ECG and EEG signals were imported

and processed in MATLAB (R2019B).

The Pan-Tompkins algorithm(Pan and Tompkins,

1985) was applied to the ECG traces to detect the R

peaks and obtain the RR time series. Such a well-

known algorithm comprises an initial band-pass finite

impulse response (FIR) filter to reduce baseline

wander and high-frequency noise, followed by

additional steps to enhance the QRS complexes and

enable R peak identification. After finding the

approximate location of the R peaks on the filtered

signals, their exact positions were determined by

finding the closest local maxima on the original ECG

signals. Finally, the identified R peaks were visually

inspected to avoid false detections or missed beats.

For each stimulation block (i.e., ten-picture

sequence), the mean RR interval and the related

standard deviation were extracted. The entire duration

of each stimulation block (2 minutes) was considered

for the calculation of these features, given the

temporal requirements for accurate estimation of

time-domain HR indices (Castaldo et al., 2019).

EEG signals were pre-processed using the

EEGLAB toolbox [https://eeglab.org/] and custom

scripts optimized for the study aim (Cassani et al.,

2022; Coelli et al., 2024). First, data were band-pass

filtered between 1 Hz and 45 Hz with a FIR (order =

400), zero-phase filter, and bad channels were

visually selected and removed (i.e., low signal quality

due to low electrode adherence).

Signals were segmented into epochs from -1 to +4

seconds with respect to each stimulus presentation.

The extended Infomax independent component

analysis (ICA) algorithm was applied to the

concatenated epochs and, with the support of the

IClabel plugin (Pion-Tonachini et al., 2019), the

sources of artifacts (i.e., ocular, heart, muscular and

residual power line noises) were identified and

removed. The previously rejected bad channels (if

any) were interpolated using spherical spline

interpolation, and signals were re-referenced to the

infinite reference using the Reference Electrode

Standardization Technique (REST plugin) (Yao,

2001). Finally, epochs with residual artifacts were

visually checked and rejected, obtaining 9.14 ± 1.07

valid trials for each participant and condition.

ERPs were obtained for each of the nine

conditions of stimulation at each EEG channel by

applying the synchronous averaging method and

using the 100 ms preceding the stimulus presentation

for baseline correction. ERP components were

extracted using a fixed time window, as suggested in

the literature (Schindler et al., 2022; Schupp et al.,

2012). We further explored specific ERP components

that have been previously correlated with emotion

perception and processing: P300 [270 - 340 ms], EPN

[200-300 ms], and the early LPP [400-700 ms].

Specifically, we identified the peak amplitude for the

Assessing Emotion-Induced Variations of Event-Related Potentials and Heart Rate During Affective Picture Processing

669

P300 and the area under the curve (AUC) for both

EPN and LPP, since these latter components cannot

be described by considering individual peaks alone.

Given that the polarity of the ERP depends on the

channel position, both the peak amplitudes and the

AUC were computed by maintaining the original

polarity. In fact, negative AUC may be obtained

because the areas corresponding to the downstate of

the ERP (negative polarity) are subtracted from the

positive one (if present). The extracted ERP and

parameters were averaged on clusters of channels:

Centrals (C3-Cz-C4), Posteriors (P7-P8-O1-O2),

Frontal Left (FP1-F1-F3), and Frontal Right (FP2-F2-

F4).

Statistical analysis was performed in R to

compare the variations in HR and ERP indices across

the different stimulation blocks.

3 RESULTS

3.1 Heart Rate Analysis Results

HR variations were analyzed considering the mean

(RRmean) and standard deviation (RRstd) of the RR

intervals. Since most of the data distributions did not

pass the Shapiro-Wilk normality test, the non-

parametric Friedman’s test and Wilcoxon’s

Bonferroni-corrected post hoc tests were used to

compare the RRmean and RRstd values among the

three valence and arousal levels separately (Figure 2).

In fact, we did not find any evidence of an interaction

between the two factors (Valence*Arousal).

Friedman’s test found significant differences for the

RRmean values across valence levels (p =0.027),

while no significant differences were found for the

Arousal main effect. The significant effect of Valence

was further explored through pairwise comparisons,

which highlighted a significant difference between

the positive stimulation against the neutral one (p

=0.008) and between the negative and neutral

stimulations (p =0.008).

Conversely, for the RRstd values, Friedman’s test

found a significant difference when comparing the

arousal levels (p =0.004) but not for the valence ones.

Specifically, HA stimuli significantly differed from

the LA ones (p =0.019). Observing an increasing

trend of the RRstd from LA to HA, we cannot exclude

an influence of the protocol sequence on this finding,

given that the IAPS pictures were presented in

increasing order of arousal.

3.2 Event Related Potential Results

Figure 3 displays the grand average of the ERP time

course for each cluster of channels (rows) grouped by

arousal level (columns). Valence levels are directly

compared in each subplot.

Since all the computed ERP indices were found to

be approximately normally distributed, a repeated

measures analysis of variance (ANOVA) with two

within factors (Valence and Arousal) was performed

for each cluster of channels. The ANOVA was

followed by post hoc tests to evaluate simple main

effects when needed.

3.2.1 P300

P300 peak amplitude results are shown in Figure 4. A

significant interaction Valence*Arousal was found in

the Frontal-Right (F(4,116) = 3.391; p = 0.012) and

Frontal-Left clusters (F(4,116) = 4.17; p = 0.003). On

the right hemisphere, significantly larger peaks were

detected in the HA block for positive and negative

valence with respect to neutral stimuli (p

Neut vs

Pos

=0.005, p

Neut vs Neg

=0.0002), and differences were

observed between HA and MA when the valence was

neutral (p

HA vs MA

=0.008). The same significant

differences were found in the left hemisphere during

HA stimulation (p

Neut vs Pos

=0.005, p

Neut vs Neg

=0.002),

and between HA and LA with neutral stimuli (p

HA vs

=0.007).

For the posterior cluster, results followed the same

pattern: a significant interaction was found

Figure 2: RR mean and standard deviation parameters compared by valence (colors) and arousal (black) levels.

Valence Arousal

BIOSIGNALS 2024 - 17th International Conference on Bio-inspired Systems and Signal Processing

670

Figure 3: ERP time course averaged across clusters of channels and volunteers for each arousal/valence condition. Solid lines

represent the group mean and shaded areas display the standard deviation.

(F(4,116) = 4.794; p = 0.001), and, dividing by

factors, we obtained significantly higher peaks in the

HA block for both positive and negative valence with

respect to neutral images (p

Neut vs Pos

=0.004, p

Neut vs

Neg

=0.001) and, at neutral valence level, between HA

and LA (p

HA vs LA

=0.0002). Finally, in the central

cluster, the interaction was again significant

(F(4,116) = 2.653; p = 0.037), and stronger P300 were

elicited in the HA block for the negative and positive

valence with respect to the neutral (p

Neut vs Pos

=0.002,

Neut vs Neg

=0.0006), and between arousal levels when

the valence was neutral (p

HA vs MA

=0.029,

HA vs

=0.002).

3.2.2 EPN

No significant interactions between Valence and

Arousal factors were identified in any cluster of

channels. Only the main effect of Arousal was found

significant at the posterior site (F(2,58) = 4.183;

p = 0.02), and the post hoc analysis identified

differences between LA and MA and between LA and

HA (p

LA vs MA

=0.014, p

HA vs LA

=0.024).

3.2.3 LPP

Distributions of the LPP area values are displayed in

Figure 5.

In the left frontal lobe, the two-factor interaction

was significant (F(4,116) =3.747; p = 0.007), and,

breaking down by Arousal levels, we found that the

positive stimuli elicited a stronger LPP with respect

to neutral stimulation (p

Neut vs Pos

=0.003) in the HA

condition. While analyzing the Valence simple main

effects, a difference was found in the positive

stimulation between HA and MA (p

HA vs MA

=0.032),

whereas HA was different from LA when the

stimulation was neutral (p

HA vs LA

=0.025). In the right

frontal lobe, only the main effect of Arousal was

found significant (F(2,58) = 3.19; p = 0.048);

precisely, LA was different from MA (p = 0.034).

At posterior sites, only the main effect of Arousal

was significant (F(2,58) = 3.173; p = 0.049) with a

significant pairwise difference between MA and LA

MA vs LA

=0.01). Finally, in the central cluster, the

interaction was again significant (F(4,116) = 3.142; p

= 0.017), and larger LPP were elicited in the HA

block for the negative and positive valence with

respect to the neutral (p

Neut vs Pos

<0.0001, p

Neut vs

Neg

=0.001) and between arousal levels when the

valence was neutral (p

HA vs MA

=0.016,

HA vs

=0.007).

Assessing Emotion-Induced Variations of Event-Related Potentials and Heart Rate During Affective Picture Processing

671

Figure 4: Comparison of P300 amplitude values across valence (colours) and arousal levels (LA, MA, HA). Pairs of * indicate

significant differences between valence levels at the corresponding arousal block (e.g., *blue-*red: difference between

positive and neutral stimulation). The coloured lines indicate the presence of a significant difference between arousal blocks

at the corresponding valence level (e.g., a blue line marks differences between arousal blocks at neutral valence).

4 DISCUSSION

In this work, we presented a classical ERP

components analysis applied to an unusual

stimulation protocol, justified by our intention to also

evaluate the slower response of the ANS through the

analysis of the RR series. Moreover, the long picture

exposure allowed us to observe the cognitive

processing of the presented stimuli (Olofsson et al.,

2008) and the associated modulations of both CNS

and ANS.

Indeed, a first interesting result was the

achievement of an expected cortical stimulation

provided by the protocol and demonstrated by the

clear ERP patterns obtained for each stimulation

condition in the different cortical regions, as defined

by clusters of channels. Specifically, ERP waves were

found prominent at posterior positions, indicating a

primary response of the visual cortex and an early

processing of the stimulus at posterior sites (Schupp

et al., 2012). Clear ERP patterns were also observed

in the frontal regions where effective content is

supposed to be further processed (Gable et al., 2014).

The interaction between the arousal and valence

emotional dimensions was significant in most cases,

particularly highlighting differences during the HA

block. Specifically, P300 and LPP were significantly

modulated by the valence of the pictures only when

the arousal level was high. In line with the literature

(Gable et al., 2014; Schindler et al., 2022; Schupp et

al., 2012), we found a stronger response to both

positive and negative pictures with respect to neutral

ones, but these were not different between them.

Unexpectedly, we found that neutral valence

images elicited weaker ERP responses in the HA

condition when compared to lower arousal levels.

This outcome might be explained by the arousal-

increasing sequence that could result in a habituation

effect, not affecting, interestingly, the volunteers’

perception of positive and negative emotions.

Moreover, such a perception was not influenced by

the presentation sequence, likely thanks to the

randomization of the different valence sub-blocks.

The EPN is often identified as a negative deflection

over fronto-central sites and a positive waveform at

lateral and posterior channels, principally modulated

by the arousal dimension of stimulating images

(Olofsson et al., 2008). Indeed, our results confirm

this statement as we coherently found a modulation

of the EPN at posterior sites induced by the arousal

dimension as a main effect.

BIOSIGNALS 2024 - 17th International Conference on Bio-inspired Systems and Signal Processing

672

Figure 5: Comparison of LPP areas across valence (colours) and arousal levels (LA, MA, HA). Pairs of * indicate the

significant differences between valence levels at the corresponding arousal block (e.g., *blue-*red: difference between

positive and neutral stimulation). The coloured lines indicate the presence of a significant difference between arousal blocks

at specific valence level. Black lines indicate significant differences between arousal levels when only the main effect of

Arousal was found significant.

In line with the EEG results, the analysis of the

HR parameters highlighted a modulation of the mean

RR due to the valence of the stimulation, resulting in

a shorter RR interval during both negative and

positive stimuli with respect to neutral ones. The

arousal dimension did not modulate the mean RR but

its standard deviation, which increased with the

arousal level. While the results related to the mean

RR are comparable with those from previous studies,

where HR features have been found more sensitive to

valence variations rather than arousal(Bensafi, 2002;

Colomer Granero et al., 2016), changes in RRstd with

arousal are not well documented in the literature.

Since the picture arousal levels increased over time in

our stimulation protocol, we cannot exclude RRstd

changes to be an effect of time rather than arousal-

dependent. In this sense, even if the protocol was

designed with an increasing arousal stimulation to

keep the subject engaged, the lack of randomization

in the arousal dimension might be considered a

limitation of the study. However, a more detailed

analysis of the HR data (e.g., including frequency-

domain features) is needed to shed light on this

peculiar finding. Moreover, to increase the statistical

power of our findings, a larger sample size is needed,

while to assess the effectiveness of the stimulation, in

terms of elicited emotions, the participants’ feedback

should be collected through questionnaires, and the

agreement among them investigated.

5 CONCLUSIONS

In summary, this study confirmed the efficacy of the

proposed stimulation protocol as it induced HR

variations and ERP responses with similar

characteristics to those observed in previous works

through different affective stimulation protocols.

Thanks to the longer exposure times guaranteed by

our protocol, these physiological responses could be

assessed both at the CNS and ANS levels through the

simultaneous analysis of ERP and HR parameters,

respectively. In this work, we focused on a time-

domain analysis, both for the cortical and ANS

response. Future studies will also include frequency

domain analysis to find optimal combinations of EEG

and HR features to predict arousal and valence levels

from physiological signals.

ACKNOWLEDGEMENTS

The study was supported by “MUSA - Multilayered

Urban Sustainability Action” project, funded by the

Assessing Emotion-Induced Variations of Event-Related Potentials and Heart Rate During Affective Picture Processing

673

European Union – NextGeneration EU (National

Recovery and Resilience Plan).

REFERENCES

Bensafi, M., 2002. Autonomic Nervous System Responses to

Odours: the Role of Pleasantness and Arousal. Chem.

Senses 27, 703–709. https://doi.org/10.1093/chemse/2

7.8.703

Cassani, C.M., Coelli, S., Calcagno, A., Temporiti, F.,

Mandaresu, S., Gatti, R., Galli, M., Bianchi, A.M., 2022.

Selecting a pre-processing pipeline for the analysis of

EEG event-related rhythms modulation. Annu. Int. Conf.

IEEE Eng. Med. Biol. Soc. IEEE Eng. Med. Biol. Soc.

Annu. Int. Conf. 2022, 4044–4047. https://doi.org/

10.1109/EMBC48229.2022.9871394

Castaldo, R., Montesinos, L., Melillo, P., James, C., Pecchia,

L., 2019. Ultra-short term HRV features as surrogates of

short term HRV: a case study on mental stress detection

in real life. BMC Med. Inform. Decis. Mak. 19, 12.

https://doi.org/10.1186/s12911-019-0742-y

Coan, J.A., Allen, J.J.B., 2004. Frontal EEG asymmetry as a

moderator and mediator of emotion. Biol. Psychol. 67, 7–

50. https://doi.org/10.1016/j.biopsycho.2004.03.002

Coelli, S., Calcagno, A., Cassani, C.M., Temporiti, F., Reali,

P., Gatti, R., Galli, M., Bianchi, A.M., 2024. Selecting

methods for a modular EEG pre-processing pipeline: An

objective comparison. Biomed. Signal Process. Control

90, 105830. https://doi.org/10.1016/ j.bspc.2023.105830

Colomer Granero, A., Fuentes-Hurtado, F., Naranjo Ornedo,

V., Guixeres Provinciale, J., Ausín, J.M., Alcañiz Raya,

M., 2016. A Comparison of Physiological Signal

Analysis Techniques and Classifiers for Automatic

Emotional Evaluation of Audiovisual Contents. Front.

Comput. Neurosci. 10. https://doi.org/10.3389/fncom.20

16.00074

Egger, M., Ley, M., Hanke, S., 2019. Emotion Recognition

from Physiological Signal Analysis: A Review. Electron.

Notes Theor. Comput. Sci. 343, 35–55.

https://doi.org/10.1016/j.entcs.2019.04.009

Gable, P.A., Adams, D.L., Proudfit, G.H., 2014. Transient

tasks and enduring emotions: the impacts of affective

content, task relevance, and picture duration on the

sustained late positive potential. Cogn. Affect. Behav.

Neurosci. 15, 45–54. https://doi.org/10.3758/s13415-

014-0313-8

Kop, W.J., Synowski, S.J., Newell, M.E., Schmidt, L.A.,

Waldstein, S.R., Fox, N.A., 2011. Autonomic nervous

system reactivity to positive and negative mood

induction: The role of acute psychological responses and

frontal electrocortical activity. Biol. Psychol. 86, 230–

238. https://doi.org/10.1016/j.biopsycho.2010.12. 003

Lang Bradley, M.M., & Cuthbert, B.N., P.J., 2008.

International affective picture system (IAPS): Affective

ratings of pictures and instruction manual.

Mizuno-Matsumoto, Y., Inoguchi, Y., Carpels, S.M.A.,

Muramatsu, A., Yamamoto, Y., 2020. Cerebral cortex

and autonomic nervous system responses during

emotional memory processing. PLoS One 15, 1–15.

https://doi.org/10.1371/journal.pone.0229890

Olofsson, J.K., Nordin, S., Sequeira, H., Polich, J., 2008.

Affective picture processing: An integrative review of

ERP findings. Biol. Psychol. 77, 247–265.

https://doi.org/10.1016/j.biopsycho.2007.11.006

Pan, J., Tompkins, W.J., 1985. A simple real-time QRS

detection algorithm A Real-Time QRS Detection

Algorithm. IEEE Trans. Biomed. Eng. https://doi.org/

10.1109/IEMBS.1996.647473

Pion-Tonachini, L., Kreutz-Delgado, K., Makeig, S., 2019.

ICLabel: An automated electroencephalographic

independent component classifier, dataset, and website.

Neuroimage 198, 181–197. https://doi.org/10.1016/

j.neuroimage.2019.05.026

Polo, E.M., Farabbi, A., Mollura, M., Paglialonga, A.,

Mainardi, L., Barbieri, R., 2023. Comparative

assessment of physiological responses to emotional

elicitation by auditory and visual stimuli. IEEE J. Transl.

Eng. Heal. Med. 1–1. https://doi.org/10.1109/

JTEHM.2023.3324249

Reali, P., Cosentini, C., Carvalho, P. De, Traver, V., Bianchi,

A.M., 2018a. Towards the development of physiological

models for emotions evaluation*, in: 2018 40th Annual

International Conference of the IEEE Engineering in

Medicine and Biology Society (EMBC). IEEE, pp. 110–

113. https://doi.org/10.1109/EMBC.20 18.8512236

Reali, P., Martinez-millana, A., Carvalho, P. De, Bianchi,

A.M., 2018b. Cardiovascular effects of stress and

emotions : a brief overview of concepts and assessment

methods 165–173.

Ruiz-Padial, E., Mercado, F., 2021. In exogenous attention,

time is the clue: Brain and heart interactions to survive

threatening stimuli. PLoS One 16, 1–20. https://doi.org/

10.1371/journal.pone.0243117

Russell, J.A., 2003. Core affect and the psychological

construction of emotion. Psychol. Rev. 110, 145–172.

https://doi.org/10.1037/0033-295X.110.1.145

Schindler, S., Bruchmann, M., Straube, T., 2022. Feature-

based attention interacts with emotional picture content

during mid-latency and late ERP processing stages. Biol.

Psychol. 170, 108310. https://doi.org/10.1016/

j.biopsycho.2022.108310

Schupp, H.T., Schmälzle, R., Flaisch, T., Weike, A.I.,

Hamm, A.O., 2012. Affective picture processing as a

function of preceding picture valence: An ERP analysis.

Biol. Psychol. 91, 81–87. https://doi.org/

10.1016/j.biopsycho.2012.04.006

Telles, S., Singh, D., Naveen, K. V., Pailoor, S., Singh, N.,

Pathak, S., 2019. P300 and Heart Rate Variability

Recorded Simultaneously in Meditation. Clin. EEG

Neurosci. 50, 161–171. https://doi.org/10.1177/155005

9418790717

Yao, D., 2001. A method to standardize a reference of scalp

EEG recordings to a point at infinity. Physiol. Meas. 22,

693–711. https://doi.org/10.1088/0967-3334/22/4/305

BIOSIGNALS 2024 - 17th International Conference on Bio-inspired Systems and Signal Processing

674