LumaDreams: Designing Positive Dream Meaning-Making for Daily Empowerment

Research Background and Issues

• Identified Problems or Challenges:

  1. Dreams significantly impact cognitive and emotional health, yet there is insufficient research and development of daily tools for engaging with dreams, especially in non-clinical settings.
  2. Current technologies primarily focus on influencing dreams unconsciously during sleep, which may undermine user autonomy and raise potential ethical concerns.
  3. Existing dream journaling tools and dream reconstruction systems are highly passive, lacking opportunities for users to actively interact with their dreams, and are mostly limited to recording and understanding.
  4. Dream transformation techniques, such as Image Rehearsal Therapy (IRT), are confined to clinical environments and are not suitable for everyday use.

• Significance:

  1. Dreams can profoundly influence self-awareness, cognitive and emotional states, and overall well-being.
  2. Positive meaning-making from dreams can aid emotional regulation and cognitive enhancement, helping users better cope with real-life challenges.
  3. Introducing dream research into the field of Human-Computer Interaction (HCI) represents a cutting-edge exploration with both innovative and practical value.

• Research Motivation and Related Work:

  1. Previous studies have used sensing technologies to influence dreams, but they face the issue of insufficient user awareness.
  2. Recent digital explorations of dream journaling indicate that recording and reviewing dreams can enhance self-understanding and emotional awareness.
  3. In psychotherapy, guided dream transformation methods have proven effective in treating chronic nightmares, among other contexts. However, these methods lack the technological support needed for everyday, widespread application.

Proposed Solution

• Proposed Method or Solution: The authors designed and developed LumaDreams, a mobile application based on generative artificial intelligence for dream journaling, visualization, scenario transformation, and positive meaning reflection.

  • Users can record their dreams in text and sketch formats.
  • The application uses AI to visualize dreams as vivid images and transform them into warmer, more positive, and healing scenarios, accompanied by related positive narratives.
  • Users can review and reflect on these "transformed dreams" to achieve daily emotional and cognitive enhancement.

• Innovative Aspects of the Solution:

  1. Shifting from passive reflection to active participation, positioning users as co-creators of dream meaning.
  2. Introducing generative artificial intelligence (Generative AI) for extracting dream content, generating visual images, and constructing positive global meanings.
  3. Emphasizing human-computer collaboration, allowing users to iteratively adjust sketches and descriptions to enhance the precision and personalization of dream transformation.

• Implementation Steps and Key Technologies:

  1. Users record dreams using multimodal inputs (sketches + text).
  2. AI models analyze and extract emotions, scenes, and elements from the dreams:
     • Image-text models (e.g., CLIP) generate descriptive text from sketches.
     • Text models (e.g., GPT-3.5 Turbo) interpret dream scenarios and identify "situational meanings" requiring transformation.
     • Text-to-image models (e.g., DALL-E 3) generate visual representations of the initial dream and its transformed positive version.
  3. Interactive features (e.g., "shattering glass effect" for dream transformation) and a dream browser facilitate reflection and review.

Research Outcomes

• Specific Results:

  1. Enhanced Daily Empowerment: LumaDreams significantly empowered users in terms of perception, cognition, and emotion, effectively enhancing their autonomy, self-efficacy, and emotional control.
  2. Improved Dream-Related Metrics:
     • Sleep quality improved significantly, with self-assessment scores increasing from low to high (WHO–5 rising from 10 to 19).
     • Nightmare frequency decreased, emotional tone of dreams improved, and users more frequently recorded and reviewed their dreams.
  3. System Evaluation:
     • System Usability Scale (SUS): Scored 77.3 (indicating a "good" level).
     • User Experience Questionnaire Short Version (UEQ-S): Rated "excellent" in both functionality and experience.

• Advantages Over Existing Solutions:

  1. Extends dream transformation techniques from clinical therapy to everyday use, greatly enhancing applicability and accessibility.
  2. Achieves more efficient and user-friendly dream visualization and meaning transformation through generative artificial intelligence.
  3. System design emphasizes positive guidance, user privacy, and cultural adaptability, addressing common ethical risks associated with AI technologies.

• Experimental or Evaluation Results:

  1. In a 14-day user test involving 14 participants, significant emotional regulation and daily empowerment effects were observed as a result of dream transformation.
  2. Quantitative and qualitative analyses both demonstrated that LumaDreams excelled in promoting deeper engagement with dreams, fostering positive meaning transformation, and improving users' daily lives.

• Limitations and Future Directions:

  1. Small sample size, primarily university students, lacking comprehensive validation across cultures and demographics.
  2. Meaning-making may not always have positive effects; further research is needed on potential negative emotional impacts of long-term use.
  3. The system currently lacks deep adaptation to users' specific cultural backgrounds and personalized needs.
  4. Risks of generative AI biases (e.g., algorithmic and cultural biases) remain, necessitating more comprehensive model correction mechanisms.

  Future Directions:

  1. Expand to larger and more diverse participant groups to validate cross-cultural adaptability.
  2. Develop personalized models deeply integrated with users' backgrounds to enhance experience and effectiveness.
  3. Explore long-term usage effects to evaluate its broader impact on mental health and life satisfaction.
  4. Optimize ethical mechanisms to ensure greater transparency and responsibility in handling sensitive data and potential inducive effects of AI systems.