Uncanny valley for interactive social agents: an experimental study

In this study, we evaluate the Strange Valley theory through live human interactions with four human-like entities:

1. Maya: simple voice assistant (only human voice).

2.: Programmable child-sized humanoid robot with poseable limbs but no skin or hair.

3. Nicole: virtual human with a full virtual human type incarnation.

4. Nadine: Life-size humanoid robot complete with skin, poseable hands, and other human-like features.

Our study aims to answer the following research questions.

· Uncanny valley for interactive humanoid robots: Exploration of this theory to provide deep insight into how people’s emotions and perceptions vary for different types of human-like interactive robots. Examine how the Uncanny Valley affects the current generation of humans, who are more accustomed to advanced technology and may be more open to human-like entities.

·Use of AI for Uncanny Valley Quantification: quantification of participants’ emotional responses using multimodal surveys and analyzes of emotions and feelings (visual, audio and textual).

The results show that the four robots were perceived in different ways and the emotions expressed about them varied. Nadine was chosen as the “favorite” robot. The only trait she failed to surpass was likability, where she tied with Nao. Additionally, Maya was seen as the least anthropomorphic and animated, which is to be expected considering she’s just a voice assistant. Nao and Nicole obtained the same score on anthropomorphism and animity. However, Nao was significantly more sympathetic, indicating that a physical body can elicit a higher degree of sympathy. Nadine, the most humanoid robot, was rated as the friendliest robot, with similar results obtained through sentiment and facial expression analyses. Nao, the humanoid but toy robot, was also considered likeable and caused the most positive surprise. However, it elicited high disgust, as determined by audio data analyses, without evoking much emotion. However, while being the friendliest, Nadine generated the most sadness and was the second most feared. This fear is perhaps indicative of the “strange valley” effect. Nevertheless, with the “weird valley”, one would expect the weirdness to increase with increasing degree of anthropomorphism, which was not the case. The greatest fear was expressed towards Maya, the voice assistant with no visual features. Also, due to Nadine’s human appearance, most attendees were hesitant and fearful as they wanted to impress her. This fear can be considered good because it indicates that people want to connect with it. By extending our multimodal analysis to other cues such as pose estimation, body language cues could be taken into account to detect evidence of a “strange valley”. While Nao had a childlike, toy-like appeal, Nadine had a human-like body, which was clearly enough to increase her likability over a bodiless or virtual agent. However, not all robots were appreciated. The lowest-ranked robot scored above 60/100, indicating that no “strange valley” effect was determined in this study. Additionally, the lack of a correlation between being more anthropomorphic and less likable and causing more negative emotions provides evidence against the hypothesis. Compared to previous research, the robots used in this investigation may not have caused the strange valley effect, as they were designed and built with care and consistency.

The results showed that the robots elicited different emotions, but the most anthropomorphic robots were liked the most. Moreover, all robots were liked and there was no correlation between the anthropomorphism of robots and the negative emotions they elicited. Therefore, this study, like previous studies, did not observe or support the strange valley hypothesis. This could be due to the characteristics of these specific robots or the specifics of today’s world in which humanoid and non-humanoid robots are becoming more prevalent and people are used to them. Either way, future interactive robot design should be open to creating anthropomorphic robots, while ensuring consistent design.