216
D. Zhang
success, the single-brain approach has recently been criticized, mostly for its limitations on ecological validity [8, 32, 38, 49]: the experimental paradigms of the
single-brain approach often employ ‘off-line’ social cognition tasks with abstract
stimuli, resulting in reduced generalizability of the findings. Although paradigms
of high ecological validity are always preferred, naturalistic social stimuli and realtime online social interactions pose great challenges for developing effective research
methods.
To further pursue neural mechanisms of our social brain, the hyperscanning technique has been developed and applied for neuroscience research in the past two
decades [2, 14, 23]. The hyperscanning technique aims at collecting and analyzing
neural signals from multiple persons involved in naturalistic social scenarios. The
collection of neural signals from interacting brains is believed to overcome the major
critiques on the single-brain approach: our brains should be better activated during
real social activities, rather than in isolated single-brain paradigms. To date, a variety
of brain imaging techniques has been employed in hyperscanning studies. The very
first study in 1965 by Duane and Behrendt utilized dual-EEG to explore a possible
extrasensory electroencephalographic induction between identical twins [12]. The
modern hyperscanning studies started from an fMRI study by Montague and colleagues, demonstrating a hyperlink between two persons playing a simple deception
game in simultaneously recording fMRI scanners [39]. Since then, more than 80
papers have been published (source: PubMed), using fMRI, EEG or fNIRS. While
fMRI has the best spatial resolution, EEG and fNIRS has gained increasing popularity in recent years, for their high portability and low running cost. Compared to both
fMRI and fNIRS, EEG has unique advantages for its rich temporal and spectral information. More importantly, the millisecond-scale high temporal resolution of EEG
is capable of following the fast temporal dynamics of human social activities. Nevertheless, further development of computational EEG analysis methods is required,
as EEG-based interacting-brain approach emphasizes the interplay among multiple
persons, for which the conventional single-brain based analysis methods cannot be
directly applied. While substantial progress has been made toward hyperscanningspecific analysis methods in the past decade, we are still at the early beginning of
exploring the ‘hyperlinks’ among interacting brains.
In this chapter, we review recent works on computational EEG analysis methods
for social neuroscience studies using the interacting-brain approach (by employing
the EEG-based hyperscanning technique). Our review is further divided into two
sessions, ‘social perception’ and ‘social interaction’. Whereas the ‘social perception’
section introduces these methods on perceiving naturalistic social information, the
‘social interaction’ section focuses on methods for characterizing the inter-person
social activities.
D. Zhang
success, the single-brain approach has recently been criticized, mostly for its limitations on ecological validity [8, 32, 38, 49]: the experimental paradigms of the
single-brain approach often employ ‘off-line’ social cognition tasks with abstract
stimuli, resulting in reduced generalizability of the findings. Although paradigms
of high ecological validity are always preferred, naturalistic social stimuli and realtime online social interactions pose great challenges for developing effective research
methods.
To further pursue neural mechanisms of our social brain, the hyperscanning technique has been developed and applied for neuroscience research in the past two
decades [2, 14, 23]. The hyperscanning technique aims at collecting and analyzing
neural signals from multiple persons involved in naturalistic social scenarios. The
collection of neural signals from interacting brains is believed to overcome the major
critiques on the single-brain approach: our brains should be better activated during
real social activities, rather than in isolated single-brain paradigms. To date, a variety
of brain imaging techniques has been employed in hyperscanning studies. The very
first study in 1965 by Duane and Behrendt utilized dual-EEG to explore a possible
extrasensory electroencephalographic induction between identical twins [12]. The
modern hyperscanning studies started from an fMRI study by Montague and colleagues, demonstrating a hyperlink between two persons playing a simple deception
game in simultaneously recording fMRI scanners [39]. Since then, more than 80
papers have been published (source: PubMed), using fMRI, EEG or fNIRS. While
fMRI has the best spatial resolution, EEG and fNIRS has gained increasing popularity in recent years, for their high portability and low running cost. Compared to both
fMRI and fNIRS, EEG has unique advantages for its rich temporal and spectral information. More importantly, the millisecond-scale high temporal resolution of EEG
is capable of following the fast temporal dynamics of human social activities. Nevertheless, further development of computational EEG analysis methods is required,
as EEG-based interacting-brain approach emphasizes the interplay among multiple
persons, for which the conventional single-brain based analysis methods cannot be
directly applied. While substantial progress has been made toward hyperscanningspecific analysis methods in the past decade, we are still at the early beginning of
exploring the ‘hyperlinks’ among interacting brains.
In this chapter, we review recent works on computational EEG analysis methods
for social neuroscience studies using the interacting-brain approach (by employing
the EEG-based hyperscanning technique). Our review is further divided into two
sessions, ‘social perception’ and ‘social interaction’. Whereas the ‘social perception’
section introduces these methods on perceiving naturalistic social information, the
‘social interaction’ section focuses on methods for characterizing the inter-person
social activities.