AffecTech

By Shadi Ghiasi 

The field of affective computing nowadays aims to implement the engineering approaches to link the physiological patterns to different states of emotion. A wide class of research has been oriented to identify the emotional responses by investigating the theoretical aspects of the autonomic nervous system (ANS). This system is response to maintain hemostasis by regulating major cardiovascular variables such as arterial blood pressure.

The mechanism behind the emotional elicitation in neuro-cardiovascular level: in general, when an emotional elicitation is present, the central nervous system is stimulated. This behavior is explained by the fact that the human emotions originate in the cerebral cortex. When the prefrontal cortex is activated, along with other cortical areas, proper changes in ANS are produced through two main subdivisions called the sympathetic and parasympathetic pathways.

Studying the dynamics of ANS provides us to measure physiological reactions during emotional experiences to obtain physiological correlates of emotion. The reflective ANS signs are characterised through powerful emotion-related signals such as heart rate variability (HRV), respiration (RSP), electrodermal activity (EDA), pupil size and the eye movement. The main objective of engineering approaches is to find reliable and robust features from the measured ANS signals which are able to characterise emotional states induced by auditory, olfactory, tactile or a multimodal emotional elicitation platform. For this aim two different attitudes are reported.

Applying Signal processing algorithms on the acquired signals is a standard approach which emphasises more on the data and allow the developed program to a gain an understanding of the problem based on the trainings on datasets. However, an alternative approach is to take into account the physiological and physiological characteristics integrated into a mathematical model which can better describe how the observable process is related to the physiological system.

Among the physiological signals the electrodermal activity, referred to alternations in the electrical properties of the skin has been recognised as a sensitive biomarker in the field of emotion recognition and the assessment of mental disorders. Several methodologies are developed to understand how the ANS dynamics regulates the Electrodermal activity (EDA) dynamics. These models take the a-priori psychophysiological knowledge to achieve insights into the relationship between the central state and the skin conductivity.

Another important emotional related signal is referred to variations in beat-to-beat intervals which is known as heart rate variability (HRV). The relation of HRV to the ANS system is reflected in the regulation mechanism. Nevertheless, the cardiovascular system exhibits a complex and nonstationary dynamics which are associated to nonlinear interactions occurring at the neuron and receptor levels. Therefore, to quantify the complexity of cardiovascular control nonlinear signal processing methodologies such as approximate entropy (ApEN), detrended fluctuation analysis (DFA) and recurrence plots (RP) have been recognised as successful approaches.

Apart from this class of analysis, several physiological modeling approaches are proposed to study the nonlinearity and complexity of cardiovascular system. A category of these models motivated by physiological studies are autoregressive models which are advantageous since they can provide spectral and cross spectral analysis. Another recognised technique inspired by taking into account the interactions between the sympathetic and the parasympathetic nervous system and the heart rate fluctuations is called the integral pulse frequent modulation (IPFM) which is also proposed in time varying version.

In conclusion, the main goal of either the conventional signal processing techniques or the physiological modeling methodologies is to obtain a better insight for human brain, cardiovascular system and their interactions and the mechanisms activated during emotional stimulus and are reliable tools to characterise these behaviors.

References

1) Valenza, G. and Scilingo, E.P., 2016. Autonomic nervous system dynamics for mood and emotional-state recognition. Springer.
2) Greco, A., Valenza, G. and Scilingo, E.P., Advances in Electrodermal Activity Processing with Applications for Mental Health.
3) Brennan, M., Palaniswami, M. and Kamen, P., 2002. Poincare plot interpretation using a physiological model of HRV based on a network of oscillators. American Journal of Physiology-Heart and Circulatory Physiology, 283(5), pp.H1873-H1886.
4) Mainardi, L.T., 2009. On the quantification of heart rate variability spectral parameters using time–frequency and time-varying methods. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 367(1887), pp.255-275.
5) Tabor, M., 1989. Chaos and integrability in nonlinear dynamics: an introduction. Wiley.

Author

Shadi Ghiasi is a PhD student at The University of Pisa under the supervision of Professor Enzo Pasquale Scilingo and a member of AffecTech project. Shadi’s main research is to apply advanced signal processing techniques and develop mathematical physiological modeling approaches to investigate the interactions of brain pathways and the cardiovascular system and extend these methodologies in the field of emotion recognition research. AffecTech is established with help from the Marie Sklodowska-Curie Innovative Training Network funded by European Commission Horizon 2020.