Sense of Agency and User Experience: Is There a Link?

Author

JOANNA BERGSTRÖM, JARROD KNIBBE, HENNING POHL, KASPER HORNBÆK

Keywords

User experience, agency, on-skin input

WHAT

link between sense of agency and user experience

WHY

Sense of control is considered an important quality of human–computer interaction (HCI).
It is unclear whether the objective and subjective measures relate to control in the same sense
A paradigm for objectively measuring SoA (called intentional binding) is hard to integrate in HCI experiments
- the task relies on low-level inputs
- the participants is hard to pay close attention and perform many repetitions
It is unclear what these subjective measures show about the sense control in HCI
- how do they relate to the objective measures?
- how do they relate to other measures of user experience?

HOW

Experienment 1
- Hypothesis: There is a difference in the sense of agency between on-skin and device inputs
- Add the touchpad condition to Coyle’s experiment, which allows us to compare a two-state input method similar to tapping on the skin, but on an external device
- Libet Clock Method
  - The phenomenon that the perceived time interval between intentional actions and their outcomes is shorter than the actual time interval
- Task
  - The participant sits in front of the screen looking at the clock, and is asked to report where the rotating arm on the clock was pointing when an action (input) was performed, or when an outcome (beep) was observed.
- Design
  - 3 input interfaces: Button, Skin, and Touchpad
  - 4 measurements collected: Action Baseline, Action Active, Outcome Baseline, and Outcome Active.
  - Order balance: Latin Squares
  - 30 times, 80 minutes
- Apparatus
  - Button: an Enter -key of a USB -connected numpad
  - Skin: piezo-electric contact microphone via a high-fidelity soundcard (Focusrite Scarlett 6i6 2nd Gen)
  - Touchpad: The MacBook’s touchpad
- Procedure
  - Introduction of tasks
  - Practice round for each of the four measurement types with each interface
  - View on-screen instructions telling them to record either (a) the perceived action or (b) the perceived outcome time
  - Press a foot switch to activate the cloak arm in a random position and start to rotate.
  - Fixed 250ms interval as an outcome and the participant told the experimenter the perceived time of action or outcome
- Participants
  - 24 right-handed participants (10 females, with an average age of 30.95 years)
  - Excluded 3 participants who misinterpreted experimental instructions
Experienment 2
- Hypothesis: Differences in experienced control correlate positively with the sense of agency
- to understand if and how interfaces that provide different levels of sense of control with objective measures also differ in subjective measures of the sense of control and other factors of user experience they provide.
- Participants
  - 42 participants (21 females, with a mean age of 27.10 ± SD of 4.89 years)
  - All used a keyboard daily, and 27 of them a touchpad
- Design
  - 3 input interfaces
  - Order balance
  - Questionaire
- Task
  - Play a simple game, using either the button, the skin, or the touchpad for input
  - 3 purposes
    - incorporate real-world interaction in order to measure user experiences with the input interfaces
    - be similar to that in previous experiments of sense of agency: binary selection
    - a measure of performance was needed as a control measure for the measures of user experience.
- Procedure
  - four reaction time tests
    - participants were asked to fixate on a blank window displayed on a screen until a green circle appeared in the middle of it
    - the participants have to react to circle appearances by pressing the button, tapping their arm, or tapping the touchpad as quickly as possible.
  - main task - play the game
- Questionaires
  - AttrakDiff
  - UMUX-LITE
  - NASA-TLX
  - Barlas and Kopp: measure the explicit sense of agency with one item asking participants to indicate the degree of control they have felt over the changes on the screen
  - Longo and Haggard: measure senses of ownership and agency over a virtual hand
  - Ebert and Wegner: one self-report question “Indicate how much it felt like moving the joystick caused the object on the computer screen to move” as a measure of explicit sense of agency
  - Subjective Duration Assessment
- Apparatus
  - Similar to Experiment 1

Results

Experiment 1
- ANOVA shows a significant difference between the intentional binding values with the three interfaces
- A Bonferroni corrected paired sample t-test further shows significantly higher outcome binding with Skin input compared to Button
- Replicated the significant difference Coyle’s work
- Support the idea that the difference in the sense of agency is due to using the skin as an input device.
Experiment 2
- Input Accuracy
  - ANOVA shows a significant difference between the error rates with the three interfaces
  - Showing that while taps are registered equally well between skin and touchpad, button input outperforms skin and touchpad in binary input
  - ANOVA shows a significant difference between the reaction times with the three interfaces
  - no difference in the accuracy between touchpad and skin
- Task Performance
  - ANOVA shows a significant difference between the hit rates with the three input types
  - eliminate the influence of performance on our measures of user experience for analysing differences between the touchpad and the skin
  - the better performance with button input may positively influence its user experience
- User Experiences
  - paired t-tests shows thatsense of agency was found higher for on-skin input compared to the two other input interfaces
  - experienced control seem to reflect the difference in the sense of control between the on-skin input and touchpad, but such relation was not found between on-skin input and button input.
  - Hedonic quality and ownership were the only measures where on-skin input resulted to more positive user experiences compared to both button and touchpad input
- Performance and User Experience
  - low-level performance in input as the reaction time may not affect subjective experiences
  - higher-level performance measure of hit rate appears to influence subjective experiences.

Discussion

Links from Agency to User Experience
- those low-level performance differences in reaction times are not perceived by users, while the higher-level performance of hit rates are
Is there a link, then?
- implicit sense of control influences explicit sense of control; however, differences in the perceived performance appear to moderate this influence
- The correlation may be moderated by performance
Limitations
- do not have actual, per-participant agency measures due to this setup
- inherent characteristics of button and touchpad input is difficult to mitigate
- Games likely come with particular expectations on part of the participants about control
- the moderating effects of performance can be a result of only small (yet significant) differences in binding effect