Shared procedure
How we measured PLAY
Every environment was tested the same way - one participant at a time, a tutorial environment first, then the affective environment, with questionnaires at three points and a fixed rule for what counts as success.
Study design
Each environment was evaluated in a within-subjects user study with 25 participants. Every participant experienced both the tutorial environment and the affective environment, and was compared against themselves. Sessions ran one participant at a time and lasted about 50 minutes.
The tutorial environment exists because first-time VR users feel excitement simply from using VR, and frustration from learning the controls. Neither is caused by the affective environment, so the participant's state after the tutorial is treated as their neutral baseline.
| Reading | When | Role |
|---|---|---|
| P1 | Before any VR | Resting state on arrival |
| P2 | After the tutorial environment | Control. The baseline every claim is measured against |
| P3 | After the affective environment | Shows what the environment did |
The seven parts of PLAY
PLAY is the highly pleasurable, fast-paced series of back-and-forth interactions between two or more beings, generalised to social joy. We broke that definition into seven measurable parts so that elicitation could be checked from several directions at once. The raw feeling is joy with high valence, arousal and dominance, where its distinct attributes are its social and reciprocal nature, and the ANSS gives a direct reading of the PLAY system itself.
| Part | What it captures | Instrument |
|---|---|---|
| Valence | High pleasure | SAM, valence |
| Arousal | Fast-paced | SAM, arousal |
| Dominance | Control over the exchange | SAM, dominance |
| Joy | The raw affect of PLAY | PANAS-X, Joviality |
| PLAY | Alignment with Panksepp's PLAY system | ANSS, PLAY subscale |
| Sociality | The NPC was experienced as a social being | Co-presence and social presence scales |
| Reciprocity | Back-and-forth between participant and NPC | In-session telemetry, where the log supports it |
The instruments
| Instrument | Form | Taken at |
|---|---|---|
| Self-Assessment Manikin (SAM) | Three picture rows for valence, arousal and dominance, each rated 1 to 9 | P1, P2, P3 |
| PANAS-X | 60 emotion words rated 1 to 5, grouped into 13 subscales | P1, P2, P3 |
| Affective Neuroscience State Scale (ANSS) | 40 items mapped to Panksepp's primary systems; LUST items omitted | P1, P2, P3 |
| igroup Presence Questionnaire (iPQ) | 14 items rated from minus 3 to plus 3 | P2, P3 |
| Perceived Other's Co-Presence (CP) | 11 statements rated 1 to 5 | P3 |
| Social Presence (SP) | 6 items rated 1 to 7 | P3 |
| Telemetry | Log of the participant's and the NPC's actions, recorded automatically by the environment. Every environment logged the actions that make up its own exchange (throws, hits, catches). Where the log carried timestamps it was used to test whether the participant's actions were timed in response to the NPC's, where it carried counts only, it describes how much interaction took place. Each environment's page says which. | During the affective environment |
Presence is an explanatory variable rather than a part of PLAY in that a stronger sense of being there (presence) makes an environment's emotional effects more impactful.
Participants and apparatus
Participants were recruited through course WhatsApp groups and direct messages at UCT. We excluded anyone under 18, anyone with an existing psychiatric disorder or on chronic stimulant medication, anyone scoring 20 or above on the Beck Depression Inventory, and anyone with a history of motion sickness, because each of these changes emotional reactivity, reduces social play, or predicts cybersickness. Ethics clearance was granted by the UCT Inter-Faculty Research Ethics Committee, reference IFHREC/02979/2026.
All three environments were tested on the same equipment, being a 2018 HTC Vive Pro headset with two controllers in the dedicated VR room in the UCT Computer Science building, driven by a Windows 10 desktop with a six-core 2.8 GHz Intel i5, 16 GB of RAM and an NVIDIA GTX 1070. The tutorial and affective environments were built in Unity 6. Questionnaires were completed on paper and captured afterwards in Excel.
The rule for success
Every environment was tested with the same three kinds of claim, stated in advance. Each paper formalises them in its own way and lists its own hypotheses and the environment pages give each set and its outcome.
| Kind of test | What it asks |
|---|---|
| Threshold at P3 | After the affective environment, is the reading high? Valence, arousal and dominance are tested against 6.33, the top third of the SAM scale; presence against its midpoint of 0; co-presence against its midpoint of 3; social presence against its midpoint of 4. |
| Change from P2 to P3 | Did the reading rise between the post-tutorial and post-affective questionnaires? Applied to the SAM dimensions, presence, and the PANAS-X and ANSS subscales. |
- The Water Balloon Fight VE and Bunny Rally VE used two-sided tests, and the Apple Interaction VE used one-sided tests, all at the 5% level. Normality is checked with the Shapiro-Wilk test and a Q-Q plot where the paired t-test is used if it passes and the Wilcoxon signed-rank test if it does not, with Holm correction across each family of comparisons.
- To claim PLAY, an environment had to meet every one of its seven parts. None did, so each environment reports which parts were met, and the project's finding is partial evidence of elicitation, with the pattern differing by environment.