Due to copyright issues, an electronic copy of the thesis must be ordered from the faculty. For the faculty to have time to process the order, the order must be received by the faculty at the latest 2 days before the public defence. Orders received later than 2 days before the defence will not be processed. After the public defence, please address any inquiries regarding the thesis to the candidate.
Trial Lecture – time and place
See Trial Lecture.
Adjudication committee
- First opponent: Senior Data Scientist, Michael Schartner, International Brain Lab, Portugal
- Second opponent: Professor Lorina Naci, Trinity College Dublin, Ireland
- Third member and chair of the evaluation committee: Professor Trygve Brauns Leergaard, University of Oslo
Chair of the Defence
Professor II Bjørn Morten Hofmann, University of Oslo
Principal Supervisor
Researcher Bjørn Erik Juel, University of Oslo
Summary
How to determine if someone or something is conscious? How to engineer an artificial consciousness? To answer these questions one must first identify candidate behavioral, functional, and structural properties a system must have in order to be considered conscious. If we are to investigate consciousness in non-biological systems, these properties must be substrate- independent.
A candidate substrate-independent property is complexity; requiring a system to possess sufficient diversity in its input-output relationships, intricacy of its architecture, and richness in its internal dynamics. Such a system is likely to exist in a state close to criticality, a state in which the system has a maximum of information processing capacity. However, the empirical results underlying this story need to be validated as measures of complexity may not behave as expected in certain conditions or systems.
In this thesis, I summarize work that tests and compares proposed measures of consciousness in conditions like general anesthesia, cognitive and attentional load, sedation of one hemisphere, and in idealized systems such as Markov models. The studies presented here generally find that measures of signal complexity behave as expected although with some surprises. The findings are presented in context of several studies that observe that states associated with unconsciousness yield a high rate of dreams—a form of conscious experience—which is not reflected by proposed measures of consciousness.
Finally, I discuss how to reconcile a high rate of dream recall in conditions often associated with unconsciousness, and the relative change in system dynamics captured by measures of complexity. In addition, the assumption that deep sedation and deep sleep are largely “unconscious” states is questioned.
The thesis work is presented against a large backdrop of philosophical and empirical work in the wider field of consciousness science, and I discuss what would really count as evidence for or against inferring consciousness in non-human or artificial systems, and whether consciousness science as currently done is on the right path to potentially contributing to the question.
Additional information
Contact the research support staff.