Research Repository

Improving Group Decision Making with Collaborative Brain-Computer Interfaces

Valeriani, D (2017) Improving Group Decision Making with Collaborative Brain-Computer Interfaces. UNSPECIFIED thesis, UNSPECIFIED.

[img] Text
thesis.pdf
Restricted to Repository staff only until 20 July 2020.

Download (42MB) | Request a copy

Abstract

Groups are generally superior to individuals in making decisions. However, time constraints and authoritarian leaders could nullify the potential advantages provided by groups. This thesis proposes a hybrid collaborative Brain-Computer Interface (cBCI) for improving performance in group decision-making. Neural signals recorded via electroencephalography are integrated with other physiological and behavioural measures to predict the likelihood of the user being correct in a decision, i.e., decision confidence. Behavioural responses from multiple users are then weighed according to these confidence estimates to obtain group decisions. The proposed cBCI has been tested with a variety of decision-making tasks, including visual matching, visual search with traditional and realistic stimuli, face recognition from multiple viewpoints, and speech perception. Groups assisted by the cBCI were significantly superior in making decisions than both individuals and traditional equally-sized groups making decisions using the majority method. This thesis also investigates the impact that a constrained form of communication has on individual and group performance in a visual-search experiment. When decision makers are able to exchange information during the experiment, their performance dramatically decreases. However, the cBCI yields superior group decisions even in this context. The confidence estimated by the cBCI is also a more reliable predictor of correctness than the confidence reported by participants after making a decision. When group members were allowed to communicate during visual search, their reported confidence was totally unrelated to the decision correctness, while in a speech perception task reported confidences were very good predictors of correctness. On the contrary, the cBCI?s confidence estimates correlated with correctness in all experiments. When critical decisions involving substantial risks have to be made (e.g., in defence), the proposed cBCI could be a useful tool to reduce the number of erroneous group decisions, thereby saving money and lives.

Item Type: Thesis (UNSPECIFIED)
Subjects: B Philosophy. Psychology. Religion > BF Psychology
Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Science and Health > Computer Science and Electronic Engineering, School of
Depositing User: Davide Valeriani
Date Deposited: 27 Jul 2017 16:05
Last Modified: 17 Aug 2017 17:16
URI: http://repository.essex.ac.uk/id/eprint/19981

Actions (login required)

View Item View Item