Are You Human? Investigating the Perceptions and Evaluations of Virtual Versus Human Instagram Influencers

Title of the Paper

Are You Human? Investigating the Perceptions and Evaluations of Virtual Versus Human Instagram Influencers

Paper Information

• Research Domain: Human-Computer Interaction, Social Media Marketing, Virtual Influencer Studies
• Keywords: Virtual Influencers, Human Likeness, Social Media, Trust, Uncanniness, EEG

Research Background and Problem

• Problem or Challenge: With the rise of virtual influencers (VIs) on social media, there is ongoing debate regarding the differences in trustworthiness and uncanniness between virtual influencers and real humans. Despite their increasingly human-like appearance, virtual influencers still face disadvantages in trust evaluations and behavioral responses from users.

• Significance: Social media has become a critical channel for global brand marketing. The widespread use of virtual influencers not only impacts consumer purchasing decisions but also raises concerns about transparency and ethical issues.

• Research Motivation and Related Work:

  • Based on the "Uncanny Valley Effect," this study explores the mechanisms influencing trust and behavioral intentions.
  • Existing research has focused on the design of social robots and virtual avatars and their impact on user perception, but it remains unclear whether AI-generated virtual influencers can achieve trust levels comparable to real humans.
  • Developing experimental methods to reveal how virtual influencers affect users' emotions and evaluations at the neurophysiological level.

Solution

• Method or Solution:

  • Research Objective: Using behavioral questionnaires and EEG studies, compare the differences between virtual and human influencers in trust, uncanniness, and behavioral intentions, while uncovering the associated neural mechanisms.
  • Research Design:
    1. Questionnaire Survey (Study 1): 150 participants provided subjective ratings on trust, uncanniness, and recommendation intentions for Instagram posts by virtual and human influencers.
    2. EEG Experiment (Study 2): Recorded EEG data from 22 participants while viewing and evaluating posts from human and virtual influencers, focusing on differences in event-related potentials (ERP).

• Innovative Aspects of the Solution:

  • First study to combine behavioral data analysis with EEG signals to explore the neural mechanisms underlying perceptions of virtual influencers.
  • Utilized N400 and LPP potential signals as potential indicators to assess uncanniness, trust, and behavioral intentions toward virtual influencers.

• Implementation Steps and Techniques:

  1. Collected Instagram posts, controlling for color and brightness to ensure fair comparisons between virtual and human influencers.
  2. Questionnaire survey employed validated Likert scales (trust, uncanniness, recommendation intention, human likeness).
  3. EEG experiment analyzed ERP components: N400 potentials (300-500ms) and LPP potentials (600-1200ms), corresponding to semantic conflict and emotional processing, respectively.

Research Findings

• Specific Findings:

  • Behavioral Data: Virtual influencers scored significantly lower than human influencers in trust ratings and recommendation intentions, while exhibiting higher levels of uncanniness.
  • EEG Signals:
    • Stronger N400 potentials were observed for virtual influencers, correlating with uncanniness and recommendation intentions.
    • Stronger LPP potentials were observed for human influencers, positively correlating with recommendation intentions.

• Comparison with Existing Solutions:

  • Provides deeper neurophysiological insights compared to purely behavioral approaches, supporting the use of N400 and LPP as potential indicators for marketing activities.

• Experimental or Evaluation Results:

  • The study corroborates previous findings regarding the lower trustworthiness of virtual influencers while revealing the relationship between different ERP signals, human likeness, and behavioral intentions.
  • EEG results demonstrate significant differences in emotional and semantic processing between virtual and human influencers.

• Limitations and Future Directions:

  • Future research could explore the impact of cultural and demographic variables on perceptions of virtual influencers.
  • Investigate whether N400 can reflect the degree of mismatch in design between virtual and human elements.
  • Conduct detailed studies on ethical issues related to virtual influencers and develop regulatory policies to enhance transparency and ethical standards.