386
V. Vouloutsi et al.
as it was reported as: “Hello”, “Trying to get attention”, “Over here!”, but also
as “goodbye”. In the case of communicate intention, the gestures in both conditions were less clear as participants thought it was about already taken a photo
or showing a photo already taken. Proxemics and participatory context could be
important to attribute the correct meaning to the gestures. Gather more people gesture in the extroverted robot was understood as “gathering”, “ushering”,
“move this way!”, however also frequently attributed to directing traffic. As the
robot in the video did not hold the camera, the context was missing, despite
mentioning this in the introduction of the online experiment. In contrast, the
introverted version caused confusion and was not properly understood. Here, in
our effort to create some sense of animacy, we may have caused excessive movement of the torso, resulting participants to attribute this behaviour as dancing
or being negative. Finally, coordinate/take the selfie for the extroverted robot
was mostly clear and understood as “come here!” “inviting someone to join”, as
well as “move over”. The introverted gesture had mixed responses: “saying no”,
“expressing negativity”, “looking for something”, “trying to find a spot for the
picture”, “saying: ‘I can’t see you”’.
Taking this feedback, we improved the robot behaviour in the cases where
the message was not clear. The overall results of the video pilot suggested that
the potential differences in perception of the behaviour of the robot as exhibited
by socSMCs may also lie in the context of the interaction and other sensorimotor
cues like proxemics.
3.3 Experiment 3 - The Selfie Task with the Robot
In total, 20 individuals (6 females, mean age 33.0 STD 9.01) participated in
groups of two in the selfie-taking task, resulting in 10 selfie interactions (5 for
the extroverted and 5 for the introverted robot). We found no correlation between
the perceived personality of the robot and the participant’s own personality in
both conditions. When looking at the average length of selfie interactions and
the average number of group selfies, no significant differences were found in
the average length of the interaction (average time for extroverted: 417 s, introverted: 356s, Z = 4.0, p = 0.345), nor the average number of group selfies taken
(extroverted: 2.4 ± 1.96 selfies, introverted: 1.2 ± 0.98 selfies, Z = 4.0, p = 0.33).
The extroverted robot successfully concluded with group (and individual) selfies
for 3 out of 5 interactions, whereas the introverted robot successfully concluded
with group selfies for 3 out of 5 interactions and one interaction where participants took individual selfies but no group selfies. Trials with the extroverted
robot tended to garner more group selfies while trials with the introverted robot
tended to garner more individual selfies. The extroverted robot resulted in a
total of 12 group selfies and 7 individual selfies while the introverted robot led
to 6 group selfies and 12 individual selfies. We found no significant differences
among conditions between the introverted and extroverted robot regarding the
Godspeed questionnaire, however, the extroverted robot scored higher in animacy compared to the introverted one, while the introverted robot scored higher
in agreeableness.
V. Vouloutsi et al.
as it was reported as: “Hello”, “Trying to get attention”, “Over here!”, but also
as “goodbye”. In the case of communicate intention, the gestures in both conditions were less clear as participants thought it was about already taken a photo
or showing a photo already taken. Proxemics and participatory context could be
important to attribute the correct meaning to the gestures. Gather more people gesture in the extroverted robot was understood as “gathering”, “ushering”,
“move this way!”, however also frequently attributed to directing traffic. As the
robot in the video did not hold the camera, the context was missing, despite
mentioning this in the introduction of the online experiment. In contrast, the
introverted version caused confusion and was not properly understood. Here, in
our effort to create some sense of animacy, we may have caused excessive movement of the torso, resulting participants to attribute this behaviour as dancing
or being negative. Finally, coordinate/take the selfie for the extroverted robot
was mostly clear and understood as “come here!” “inviting someone to join”, as
well as “move over”. The introverted gesture had mixed responses: “saying no”,
“expressing negativity”, “looking for something”, “trying to find a spot for the
picture”, “saying: ‘I can’t see you”’.
Taking this feedback, we improved the robot behaviour in the cases where
the message was not clear. The overall results of the video pilot suggested that
the potential differences in perception of the behaviour of the robot as exhibited
by socSMCs may also lie in the context of the interaction and other sensorimotor
cues like proxemics.
3.3 Experiment 3 - The Selfie Task with the Robot
In total, 20 individuals (6 females, mean age 33.0 STD 9.01) participated in
groups of two in the selfie-taking task, resulting in 10 selfie interactions (5 for
the extroverted and 5 for the introverted robot). We found no correlation between
the perceived personality of the robot and the participant’s own personality in
both conditions. When looking at the average length of selfie interactions and
the average number of group selfies, no significant differences were found in
the average length of the interaction (average time for extroverted: 417 s, introverted: 356s, Z = 4.0, p = 0.345), nor the average number of group selfies taken
(extroverted: 2.4 ± 1.96 selfies, introverted: 1.2 ± 0.98 selfies, Z = 4.0, p = 0.33).
The extroverted robot successfully concluded with group (and individual) selfies
for 3 out of 5 interactions, whereas the introverted robot successfully concluded
with group selfies for 3 out of 5 interactions and one interaction where participants took individual selfies but no group selfies. Trials with the extroverted
robot tended to garner more group selfies while trials with the introverted robot
tended to garner more individual selfies. The extroverted robot resulted in a
total of 12 group selfies and 7 individual selfies while the introverted robot led
to 6 group selfies and 12 individual selfies. We found no significant differences
among conditions between the introverted and extroverted robot regarding the
Godspeed questionnaire, however, the extroverted robot scored higher in animacy compared to the introverted one, while the introverted robot scored higher
in agreeableness.