Virtual avatars & Immersive education

seminar 3

Immersive education & VR

• VR provides realistic scenarios allowing learning of abstract or inaccessible topics

• many studies found immersive learning beneficial

• however needs to be further explored

Immersive education

• not a single pedagogy or specific practice

• wide range of approaches to teaching & learning

• interdisciplinary & collaborative framework

• commonly associated w/VR & active learning

Gardner cycle

• Each new technology goes through a cycle 

1. technology trigger

2. peak inflated expectations

• excited about research

3. trough of disillusionment

• realising the actual effect

4. slope of enlightenment

5. platuea of productivity

edify teacher

• interact & set up lesson & broadcast

• teacher is 3d presenting it to the students

• students access and interact remotely 

• moslty during covid

multi user VR

• when you are all in the same VR place but across the world

• like frame VR

bay stations/play area

• are in which an individual can use VR to play & track 

• hard to space out 

• requires limited things on the floor

Exemplar apps

• Virtual printing press

  • students learning the history of printed by doing in VR

Preliminary user testing of project mobius

• some apps were explored more than others e.g disease diagnosis

• instructions were followed easily w/minor errors

• controllers were used w/minor errors

• good feedback

Data collection (Savickaite et al 202?)

• The initial unique data which is available is the student’s position in physical and virtual space

• The following examples were collected in the Disease Diagnostics lesson

• User interaction data from the Mobius apps can be exported as a .json file and converted to .csv format for further analysis.

Since the initial mobius study

• VR became more mainstream

• no longer need for controllers

Challenges with VR & immersive learning

• clarity of instructions

• expertise

• experience of controls was hard

Accessibility Considerations in XR

• Customizable Experience

  • different filters for colourblindness

• Selective Simulation

  • can be good for empathy, but danger of stereotyping disorders

• Haptics

  • spatial audio 

• Neurodivergence

  • consider how much info is given at time

• Co-Design

  • making sure accessiblilty is throughout not just at the end 

Uncanny Valley (UV)

• First proposed in 1970s by Mori

• describes/observes negative reaction to human-looking entities such as avatars

  • manifest as a cold eerie & repellent feeling

Uncanny valley hypothesis (Mori et al, 2012)

• affinity for artificial humans rises with human-likeness up to a point, then drops sharply when the entity is almost - but not quite - human, before rising again as it becomes indistinguishable from a real person

what increases uncanny valley effect

• movement

  • mismatches in timing dynamics or micro expressions

  • feel eerie 

uncanny valley depends on (Diel et al, 2021)

• appearance, animation quality, and context

• Is not inevitable if appearance–behaviour congruence is high

Research on uncanny valley (2005+)

• 2/3 of studies found affect still exists 

• cognitive & neural basis is still unclear

• facial expressions in human like avatars affect behavioural realism

Virtual avatar

• digital representation of a user

• vary in visual fidelity (abstract/human) to behavioural (canned animations to facial capture)

• bridges gap between the digital & physical world providing sense of presence

what makes avatars effective

• realistic movement

• facial expressions

• ability to interact w/objects in a natural way 

Avatar design (Latoschik et al, 2017)

• Choices of visual fidelity to behavioural fidelity influence the presence, embodiment, social copresence, and emotional recognition during interaction 

Avatar realism

• over-riding factor which impacts both the aesthetic & functional characteristics of an avatar.

• Increased realism does not always = increased approval of the avatar

  • increasing realism in avatars intensifies the sensitivity to cues perceived by end-users highlighting falsehoods in the appearance and behaviour of the avatar

• Human perception of realism varies among individuals.

what does increasing realism do for experience of avatar (Higgins et al, 2021)

• improve ownership & social interaction 

• may trigger aversion if the appearance is almost human-like 

ALIVE project (Savickaite et al)

• basic emotions repeated 3 times (anger, disgust, fear, surprise, contempt, happiness and sadness)

• avatars 1 (Wolf3D) vs avatars 2 (MetaHuman)

• VR (Oculus Quest1)

• Measures: Ratings of emotions, Errors, Reaction Times, Demographics (gender, age, VR experience)

mystery study’s results

• Meta-humans are perceived as more 'uncanny' than the cartoons → have slightly worse animation quality, but better emotion recognition accuracy

  • Each of these varied a lot by the specific emotion

• it is likely a product of the exaggerated expressions on the meta-humans

  • more exaggerated expressions = more 'uncanny' perception rating but better recognition

• P reported that some expressions were very evident but made the avatar look strange (both cartoon & meta-human)

• darker skin tone made avatars seem more realistic

  • Some emotions were very hard to distinguish (particularly in the cartoon condition).

follow up mystery study results

• ND view metahumans as uncannier than NT

  • experience greater discomfort & UV

• ND have a lower mean accuracy than NT

• ND are less accurate w/negative emotions the NT

• Females have higher accuracy than males

• no significance between avatar race 

Limitations of ALIVE study

• Metahumans

• Self-reported questionnaire

Future research

• Different VR settings

• Exploring different avatars

• Looking at diverse populations for between races of avatar and participant

Limitation of VR avatars

• Technological difficulties

  • latency is major issue

  • scalability

• privacy & security

Creation of digital avatars

• defining avatar appearance using software for customisation

• often employ 3d modelling techniques e.g Blender

• rigged w/skeleton structure enabling movement & interaction

• can be animated using motion capture technology

limitations of creating avatars

• achieving a high level of realism without sacrificing performance

• realistic avatars require significant computational power

Metahuman

• framework/toolset for creating fully rigged high fidelity digital humans w/controllable face, body hair, clothing & materials