Skip to main content

Mechanisms and Consequences of Attributing Socialness to Artificial Agents

Periodic Reporting for period 2 - SOCIAL ROBOTS (Mechanisms and Consequences of Attributing Socialness to Artificial Agents)

Reporting period: 2018-04-01 to 2019-09-30

Understanding how we perceive and interact with others is a core challenge for social cognition research. This challenge is poised to intensify as the importance of artificial intelligence, and the presence of humanoid robots in society, grows. Through an innovative combination of psychology, neuroscience and robotics, the SOCIAL ROBOTS project prepares us for this future by (1) establishing a new approach for understanding how the brain processes and responds to humanoid robots; and (2) using this approach to delineate factors influencing how flexibly representations of robots and humans are shared at brain and behavioral levels. To achieve this, we first establish how young adults perceive and interact with humans vs. robots, the role of physical features and training experience, and the extent to which brain regions mediating social interaction with humans also support robot interaction. Next, to test the role of experience-dependent plasticity on social cognition, we assess how cognitive flexibility toward robots manifests among young children and older adults. Finally, we explore cultural influences on shared representations of humans and robots by extending the first project phase to Japan, the world’s most robotics-rich nation. Findings will provide a detailed and nuanced understanding of how brain mechanisms supporting social engagement with people are used when interacting with robots, and how different kinds of experience (e.g. training, lifespan, cultural) influence this engagement. The planned experiments and those generated during the project will establish the research team as a world-leading group bridging social cognition, neuroscience and robotics.
The project is currently at the planned stage of continued data collection for all research streams, which focus on how young adults and older adults perceive and interact with social robots compared to other people, using brain imaging, behavioural and training measures. We have completed two full neuroimaging training studies with young adult participants (each study with pre- and post-training brain scanning sessions), and a number of behavioural studies, with each of these studies progressing toward publication and one published. In addition, we presented public engagement events based on the objectives of the project, ensuring individuals beyond academia are invested in and benefit from the research undertaken.

So far, the team has produced 21 publications and 25 invited conference talks for the PI alone, as well as organised a multidisciplinary workshop on the social neurocognition of human robot interaction (Bangor University, August 2017) and a symposium on a similar topic at the 2018 HRI meeting in Chicago, USA. Moreover, two team members (Cross and Hortensius) served as co-editors of a theme issue of Philosophical Transactions of the Royal Society B along with fellow ERC awardee, Agnieszka Wykowska, titled ‘From social brains to social robots: Applying neurocognitive insights to human-robot interaction’, published in 2019.

Talk highlights
2019: Keynote - Dance Data, Cognition, and Multimodal Communication Workshop (Portugal)
Embodied Social Agents workshop (Scotland)
2018: Max Planck Institute for Empirical Aesthetics (Germany)
Center for Research on Cognition and Learning, CNRS (France)
Aegina Summer School on Social Cognition, Aegina (Greece)
ERC Conference ‘Frontier Research in Artificial Intelligence’(Belgium)
UK Government Digital Service Academy, Cabinet Office, London (England)
2017: ERC 10th Anniversary Celebration invited speaker (Belgium)
British Science Festival prize lecture presenter (England)
Whitehead Lecture on Cognition, Computation and Culture, Goldsmiths (England)
Culture & Brain Seminar Series, Karolinska Institutet, Stockholm (Sweden)
"Our project has generated exciting new insights into the longer-term impacts of human social interaction with robots, primarily from neuroscience perspectives. Our studies have been the first to use representational similarity methods (MVPA) to examine the extent to which humans might perceive artificial agents as social agents/companions (Hortensius and Cross, in prep; Cross et al., 2019). As we have just passed the half-way point of the grant, we are now in full swing with gathering data for Research Stream 1 and part of Research Stream 2, and our collaboration with colleagues in Japan for Research Stream 3 is about to intensify starting in August 2019.

The project has achieved a high level of public engagement too. Our ‘Popularity Contest at the Robotic Petting Zoo’ project was featured on the BBC and other news websites, as well as BBC Radio Wales, and in the first instalment of this interactive event, we attracted nearly 1000 visitors of all ages and backgrounds to take part over a 1-week period (December 2017). Since then, we have run the event at the Glasgow Science Festival and the British Science Festival, attracting several thousand more participants. In addition, the SOCIAL ROBOTS project was featured in the ERCcomic You, Robot, published in 2018 (see https://www.erccomics.com/comics/you-robot)

In addition, eight team members maintain active twitter accounts, with frequent tweeting of #TeamSoBots project updates. We have an active project website where we post research updates and share the papers and presentations prepared by team members and students as part of our social robotics journal club (www.so-bots.com) as well as a dedicated website for the robotic petting zoo outreach project (www.roboticpettingzoo.com). We curate an active Open Science Framework account, where we share our research materials and data with the general public (as well as other academic users) in an aim to foster the transparency and reproducibility of our team’s research. We estimate that our dissemination and communication activities have reached ~9,000 members of the general public.

In terms of expected results until the end of the project, we are naturally looking to build upon all the foundational work we have been establishing over the past 30 months to extend our understanding to developmental and aging populations, as well as develop and refine cross-cultural perspectives on the dynamic nature of human-robot interactions.

PUBLICATIONS
1. Henschel, A. & Cross, E. (in press). Interaction Studies.
2. Williams, E., Cristino, F. & Cross, E. (2019). Cognition.
3. Apšvalka, D., Cross, E. & Ramsey, R. (2019). Cognition.
4. Cross, E. &al. (2019). Phil Trans Roy Soc B.
5. Cross, E., Hortensius, R. & Wykowska, A. (2019). Phil Trans Roy Soc B
6. Apšvalka, D., Cross, E. & Ramsey, R. (2018). J Neuroscience
7. Kirsch, L. &al. (2018). Neural Plasticity.
8. Hortensius, R. & Cross, E. (2018). Annals NYAS
9. Hortensius, R., Hekele, F. & Cross, E. (2018). IEEE-TCDS
10. Kirsch, L. & Cross, E. (2018). Progress in Brain Research
11. Sumanapala, D. &al (2018). Progress in Brain Research
12. Apšvalka, D., Ramsey, R. & Cross, E. (2018). Neural Plasticity
13. Lakens, D. &al (2018). Nature Human Behaviour
14. Williams, E. & Cross, E. (2018). Cognition
15. Cross, E. &al. (2017). PLoS One
16. Gardner, T., Aglinskas, A. & Cross, E. (2017). Neuroimage
17. Sumanapala, D. &al. (2017). Acta Psychologica
18. Kirsch, L. &al. (2016). PLoS One
19. Kirsch, L., Urgesi, C. & Cross, E. (2016). Neurosci Biobehav Reviews
20. Cross, E. &al. (2016). Phil Trans Roy Soc B
21. Catmur, C., Cross, E. & Over, H. (2016). Phil Trans Roy Soc B"
sobots-nobg.png
sobots-orangeblue.png