Final Report Summary - DEFCON1 (A NEW DEFINITION OF CONSCIOUSNESS) What is consciousness? The traditional answer to that question has always been something like ‘consciousness is that I know what goes on around me’. That sounds perfectly reasonable, as it fits the difference between sleeping and being awake, or the difference between crashing into an upcoming car that you didn’t notice and seeing that car and avoiding the accident. There is just one problem with this introspective / behavioral definition of consciousness: it didn’t get us anywhere in understanding the phenomenon. Is it just us, humans, who have consciousness, or do all animals that go through sleep-wake cycles (such as cockroaches) have it? Do machines that drive cars and avoid each other have consciousness?A major problem with the traditional view of consciousness is that it relies heavily on the ability to report about conscious sensations. The psychologist will always have to pose the question (one way or another): “did you see that?” Introspectively, you will always have to ask yourself: “was I aware of that?” This poses a real problem when asking whether there consciousness in a fetus, in a patient in vegetative state or in a dog that looks you in the eye. The introspective / behavioral definition of consciousness relies heavily on reportability, cognitive access, and functions like attention, memory, or language. And hence it easily conflates consciousness with reportability, access, memory, attention and language.This project has tried to overcome these problems by posing and testing a new definition of consciousness (DEFCON1), where consciousness is defined in neural terms (assuming the brain has something to do with it). It was hypothesized that the unconscious - conscious divide could be defined as the difference between feedforward and recurrent processing in the brain. Moreover, it was hypothesized that the reason for that is that recurrent – and not feedforward - processing induces NMDA receptor activation and hence learning. A major consequence of this neural definition of consciousness would be that consciousness should be viewed as independent of cognitive functions like access and attention, and hence that there can be conscious processing without the subject knowing it. This obviously forms a strong departure from the traditional introspective intuition that ‘you (and only you) know what you are conscious of’. The experimental work focused on getting empirical evidence for these strong claims in the domain of vision, using human subjects and methods like EEG, fMRI, TMS, and pharmacological interventions.The first subproject found that consciousness is independent of (orthogonal to) cognitive functions, even those that were traditionally thought to depend on conscious processing. Both visible and invisible cues were found to activate high level cognitive modules in the brain (fMRI, EEG), and to ignite high level cognitive functions like categorization, flexible cognitive and inhibitory control, evidence accumulation and decision making. For every cognitive function there seems to be a conscious and an unconscious version, supporting the idea that consciousness and cognition are orthogonal.The second subproject showed that recurrent processing (fMRI, EEG, disrupted with TMS) mediates functions that are pivotal in understanding the phenomenal character of conscious visual sensations, such as perceptual grouping, inference and completion. The neural definition of consciousness thus enables a deeper insight into why conscious sensations ‘look the way they look’, i.e. have phenomenal qualities.The third subproject confirmed the hypothesis that unattended representations have phenomenal properties like perceptual organization, inference, and amodal completion, as was shown both by neural measurements during inattentional blindness (EEG, fMRI), as by studying (fMRI, TMS, VBM) the nature of a new form of visual memory, called ‘fragile visual short term memory’. Results warrant the conclusion that we may indeed have conscious sensations without knowing it, because there seems no difference between reported and not attended / accessed / reportable visual representations, while both differ strongly from truly unconscious (e.g. masked) representations.The forth subproject provided the first evidence that consciousness and attention should also be regarded as different with respect to their roles in perceptual learning. Perceptual learning, studied both neurally (EEG, fMRI) and behaviorally, is possible without attention yet not without conscious sensation.The fifth subproject provided some very interesting leads into the molecular mechanisms required for recurrent processing and conscious vision. Some of these are indeed intricately related to plasticity and learning (NMDA receptor pathways, studied with Ketamine). But we also found a pivotal role for GABA, both in mediating the conscious state, as in modulating the content of consciousness in bistable stimuli.All in all, the new ‘neural definition’ of consciousness has shown to provide a fruitful general framework for understanding the ontology, nature, and functions of consciousness. Moreover, it has shown how we can depart from introspection and start to have a more objective scientific stance on consciousness.
