The Use of Social Sensorimotor Contingencies in Humanoid Robots
387
We further evaluated the enjoyment of the interaction. A Mann-Whitney test
indicated no significant differences in the enjoyment of the interaction between
the two conditions. Similarly, no differences we found in the following items:
“Taking a selfie with the robot was engaging”, “Taking a selfie with the robot
was annoying”, “Taking a selfie with the robot was natural”, “I liked taking
a selfie with the robot”, “I was uncomfortable taking a selfie with the robot”.
Nonetheless, participants reported taking a selfie with the extroverted robot as
easier than the introverted one (U = 24, p = 0.0226). Finally, the habit of taking
selfies did not seem to affect the participant’s enjoyment of the interaction, as
there was no significant difference in the enjoyment of the interaction based on
participants’ normal selfie-taking behaviour. These results suggest that there was
not enough evidence to support the hypothesis that participants would prefer the
extroverted robot and that the extroverted robot would lead to a greater number
of selfies and longer interactions compared to the introverted one. Nonetheless,
participants were able to understand the behaviour of the robot as expressed by
sensorimotor information and take a selfie with the robot.
4 Discussion and Conclusions
Due to a small sample size of this pilot study, the lack of significance amongst the
two conditions is not surprising. While there is not enough evidence to support
our hypotheses, the fact that both the introverted and extroverted conditions
led to a successful coupling for a group selfie indicate that the expression of
socSMCs is a viable method for coupling in Human-Robot Interaction scenarios.
While this study provides a preliminary insight, several limitations need to be
discussed.
The robot’s behaviour was based explicitly on sensorimotor contingencies,
and the resulting behaviours were based on animations conducted by the authors.
Observational data from both the video and the debriefing session after the end
of the main experiment suggest that further work needs to be conducted for the
generation of fluent and transparent gestures. Other aspects of the experimental
design may have affected the perception of the robot and task. For example,
participants did not view the attached camera as serving an “entertainment”
purpose as we would have expected. Instead, several participants perceived the
camera as an inherent part of the robot (manufactured) that allowed the robot to
see behind itself or as an additional method implemented to monitor the experiment. Participants reported that a cell phone combined with a selfie stick would
make the task more transparent, as this is the traditional way of taking a selfie.
However, due to the mechanics of the robot’s hand and the maximum weight
it can hold, such an approach is impossible. Some participants also mentioned
that the experimental setting of the XIM led them to overthink the meaning of
the gestures, and a more natural setting might have made it easier for them to
understand the communicated message. This supports HRI research focusing on
ecologically valid settings, as many social interactions (like taking a selfie) may
be inhibited by the environmental and contextual expectations of the experimental settings.
387
We further evaluated the enjoyment of the interaction. A Mann-Whitney test
indicated no significant differences in the enjoyment of the interaction between
the two conditions. Similarly, no differences we found in the following items:
“Taking a selfie with the robot was engaging”, “Taking a selfie with the robot
was annoying”, “Taking a selfie with the robot was natural”, “I liked taking
a selfie with the robot”, “I was uncomfortable taking a selfie with the robot”.
Nonetheless, participants reported taking a selfie with the extroverted robot as
easier than the introverted one (U = 24, p = 0.0226). Finally, the habit of taking
selfies did not seem to affect the participant’s enjoyment of the interaction, as
there was no significant difference in the enjoyment of the interaction based on
participants’ normal selfie-taking behaviour. These results suggest that there was
not enough evidence to support the hypothesis that participants would prefer the
extroverted robot and that the extroverted robot would lead to a greater number
of selfies and longer interactions compared to the introverted one. Nonetheless,
participants were able to understand the behaviour of the robot as expressed by
sensorimotor information and take a selfie with the robot.
4 Discussion and Conclusions
Due to a small sample size of this pilot study, the lack of significance amongst the
two conditions is not surprising. While there is not enough evidence to support
our hypotheses, the fact that both the introverted and extroverted conditions
led to a successful coupling for a group selfie indicate that the expression of
socSMCs is a viable method for coupling in Human-Robot Interaction scenarios.
While this study provides a preliminary insight, several limitations need to be
discussed.
The robot’s behaviour was based explicitly on sensorimotor contingencies,
and the resulting behaviours were based on animations conducted by the authors.
Observational data from both the video and the debriefing session after the end
of the main experiment suggest that further work needs to be conducted for the
generation of fluent and transparent gestures. Other aspects of the experimental
design may have affected the perception of the robot and task. For example,
participants did not view the attached camera as serving an “entertainment”
purpose as we would have expected. Instead, several participants perceived the
camera as an inherent part of the robot (manufactured) that allowed the robot to
see behind itself or as an additional method implemented to monitor the experiment. Participants reported that a cell phone combined with a selfie stick would
make the task more transparent, as this is the traditional way of taking a selfie.
However, due to the mechanics of the robot’s hand and the maximum weight
it can hold, such an approach is impossible. Some participants also mentioned
that the experimental setting of the XIM led them to overthink the meaning of
the gestures, and a more natural setting might have made it easier for them to
understand the communicated message. This supports HRI research focusing on
ecologically valid settings, as many social interactions (like taking a selfie) may
be inhibited by the environmental and contextual expectations of the experimental settings.