Virtual Avatars
Research 2019
Research project I helped conduct at the University of Massachusetts Boston about users attention to virtual avatars personalities through their body motions.
My Role:
Participant screening and recruitment, Stimulus design, and Running in lab VR study
Team:
Funda Durupinar(advisor) & Biao Xie (VR Developer)
Research Objectives
Objective #1 Determine how the non-verbal personality expression of a virtual character impacts overall user attention, performance and emotional arousal; and explore the combined effects of the personalities of the character and the user on these factors
Objective #2 Determine the impact of the motion of each body part on user attention, and how this changes with the virtual character’s and the user’s personalities
Hypotheses:
-
Personalities of virtual characters manifested in face and body motion affect users’ overall attention, performance and emotional reactions in a virtual reality setting.
-
Certain character personality types are more effective to engage attention and arouse emotions.
-
Human observers’ own personalities determine their preference for agent personality.
Methods
-
First participants filled out the Ten Item Personality Inventory (TIPI)
-
Participants then completed two viewing tasks using FOVE Virtual Reality head mounted display
Task #1 Attentional Disengagement Task:
During this task, participants were asked to watch the humanoid walk in place and were instructed to turn their gaze to small spherical stimulus randomly appearing on the left or right periphery
The task consisted of three blocks with 10 trials. In each block, the humanoid was assigned a different extroversion value of -0.8, 0 and 0.8 where -1 refers to complete introversion and 1 refers to complete extroversion. The blocks were presented in random order
Task #2 Two Character Viewing:
During the second task, participants viewed two characters, one with extroverted (with a value of 0.8 E) and the other with introverted (with a value of -0.8 E) personality placed side by side, randomly assigning their placement on the left or right.
Introvert
Analysis
Extrovert
Task #1 Attentional Disengagement Task:
-
We recorded the latency between the target onset and saccade start time.
-
We then calculated the per participant median latency for the 10 trials.
-
For each personality type, we presented the stimulus 10 times throughout the task block and recorded the median response time
Task #2 Two Character Viewing:
-
We computed the overall attention dwell time on each character
To evaluate the effects of the participant’s own personality:
-
We took the time difference between the median latency in extroverted and introverted cases for Task #1, and the time difference between the overall attention in extroverted and introverted cases for Task #2. We then computed Pearson correlation coefficients between these values and the participants’ extroversion scores
Results
Results show a moderate positive correlation of 0.62 between the time difference of introvert and extrovert characters and the introversion value of TIPI (being reserved and quiet), p value = 0.055. This suggests a slight tendency for introverts to focus more on the introverted characters.
However, when these characters were shown side by side, introverts showed tendency to focus on the extroverted character more, r value = -0.54 and p value = 0.11.
None of these values were statistically significant at the 0.05 level so it is hard to make any conclusions based on these results.