Adaptive joint cognitive systems for complex and strategic decision making

Building trust in human-machine teams through brain-computer-interface augmentation, social interaction and mutual learning

Personal, everyday decisions tend to be straightforward. What should I have for lunch? Shall I take a shower? What colour shirt will I wear today?

Deciding on behalf of others, however, in a way that is both effective and fair, can be very difficult. For example, after a natural disaster such as a hurricane or earthquake, quick and difficult decisions must be made about allocation of resources. Who do you prioritise for rescue, or to receive bottled supplies of clean water? If 90% of your agricultural land has been destroyed but 80% of the injuries and deaths are in a major city, which area should be prioritised for restoring services?

Artificial intelligence can be used for simple decisions, but complex decision-making is still in its infancy. This project aims to develop a brain-computer interface (BCI) which will assist and augment human decision makers, a set of AI-based systems that are capable of learning and making independent decisions and will act as team members on par with the humans, and an AI-system that will coordinate the interactions of humans and AIs and will integrate their decisions.

The particular foci of the Essex team are the development of machine learning based AI team-members, and group-decision systems that optimally integrate member decisions. Our work involves identifying neural (EEG-based) and behavioural correlates of decision confidence, and BCIs for real-time estimates and prediction of decision confidence and user mental states, as well as identifying the behavioural, physiological and neural markers of effective group cooperation.

The end result of this work will be a prototype system for dealing with situations requiring strategic decisions, such as allocating resources in emergency situations.

Research background and expertise

The project is led by Professor Riccardo Poli in the UK, and Professor Maryam Sanechi in the US. It aims to develop a novel architecture for complex group decision-making that integrates, in an unprecedented way, the strengths of human and AI team members while compensating for their respective weaknesses.

The project involves a multidisciplinary effort, involving expertise in AI, machine learning, neural engineering, computer science, neuroscience, and cognitive psychology. It builds on many years of highly interdisciplinary research on group decision making assisted by Brain-Computer Interfaces (BCIs) in human and human-machine teams, as well as state of the art machine-learning technology, neuroscience, human-factors and psychophysiological knowledge on decision making in humans and human teams.

Funding

The project is funded via the Bilateral Academic Research Initiative (BARI), which is a pilot program designed to support high-risk basic research as a bilateral academic collaboration; it is jointly sponsored by the Basic Research Office in the US Office of the Secretary of Defense, and the UK Ministry of Defence.

Partners

BARI is a multi-centre project involving institutions and teams from both the UK and the US.

UK institutions

• University of Essex; Professor Riccardo Poli (PI), Professor Luca Citi, Dr. Caterina Cinel, Dr Tasos Papastylianou, Dr Rodrigo Ramele.
• University of Oxford; Professor Nick Yeung, Dr Matt Davidson, Sriraj Aiyer.

US institutions

• University of Southern California; Professor Maryam Shanechi (PI), Nitin Sadras (PhD student), Omid Ghasem-Sani (PhD student).
• University of California, Berkeley; Professor Michael Jordan.
• University of Massachusetts Medical School; Professor Farid Hamzei-Sichani, Dr Yuming Huang.
• Harvard Medical School; Professor Kristina Simonyan.

Publications

• Ahmadipour, Parima, Yuxiao Yang, Edward F. Chang, and Maryam M. Shanechi. “Adaptive Tracking of Human ECoG Network Dynamics.” Journal of Neural Engineering, 2020.
• Bhattacharyya S, Valeriani D, Cinel C, Citi L, Poli R. Anytime Collaborative Brain-Computer Interfaces for Enhancing Group Decision-Making in Realistic Environments. Scientific Reports (under review).
• Sani, Omid G., Hamidreza Abbaspourazad, Yan T. Wong, Bijan Pesaran, and Maryam M. Shanechi. “Modeling Behaviorally Relevant Neural Dynamics Enabled by Preferential Subspace Identification.” Nature Neuroscience 24, no. 1 (January 2021): 140–49.
• Valeriani D (2020). Neurotechnologies for Optimal Human-Machine Collaboration in Decision-Making. 2020 IEEE Brain Workshop on Advanced Neurotechnologies.
• Valeriani D, O’Flynn LC, Worthley A, Simonyan K (2020). Neural Correlates of Accuracy and Confidence during Realistic Decision-Making in Noisy Environments. Organization for Human Brain Mapping (OHBM) Annual Meeting 2020.
• Valeriani D, Poli R (2019). Cyborg groups enhance face recognition in crowded environments. PLOS ONE 14(3): e0212935.
• Yang, Yuxiao, Parima Ahmadipour, and Maryam M. Shanechi. “Adaptive Latent State Modeling of Brain Network Dynamics with Real-Time Learning Rate Optimization.” Journal of Neural Engineering, 2020.