I have posted on this before (here and here) but I am going to again, prompted by Tallon-Baudry’s talk at the Turing 2012 Consciousness Event (video). The experimental detail covered in the talk is from a paper with Wyart and Dehaene (citation below). She reviewed this area of research recently (citation below). Abstract:
Consciousness, as described in the experimental literature, is a multi-faceted phenomenon, that impinges on other well-studied concepts such as attention and control. Do consciousness and attention refer to different aspects of the same core phenomenon, or do they correspond to distinct functions? One possibility to address this question is to examine the neural mechanisms underlying consciousness and attention. If consciousness and attention pertain to the same concept, they should rely on shared neural mechanisms. Conversely, if their underlying mechanisms are distinct, then consciousness and attention should be considered as distinct entities. This paper therefore reviews neurophysiological facts arguing in favor or against a tight relationship between consciousness and attention. Three neural mechanisms that have been associated with both attention and consciousness are examined (neural amplification, involvement of the fronto-parietal network, and oscillatory synchrony), to conclude that the commonalities between attention and consciousness at the neural level may have been overestimated. Last but not least, experiments in which both attention and consciousness were probed at the neural level point toward a dissociation between the two concepts. It therefore appears from this review that consciousness and attention rely on distinct neural properties, although they can interact at the behavioral level. It is proposed that a cumulative influence model, in which attention and consciousness correspond to distinct neural mechanisms feeding a single decisional process leading to behavior, fits best with available neural and behavioral data. In this view, consciousness should not be considered as a top-level executive function but should rather be defined by its experiential properties.
She outlines three ways of relating consciousness and attention.
Gateway hypothesis: attention and consciousness reflect related concepts, much as temperature and heat. The idea that attention acts as a gateway for consciousness has been formalized in influential theories of consciousness: those events that enter consciousness are those that have been selected and amplified by attention.
Reverse dependence hypothesis: alternatively, whether a stimulus has been detected or not at the neural level could trigger different attentional mechanisms.
Cumulative influence hypothesis: attention and consciousness would be implemented by distinct neural mechanisms, but would both influence, although with different weights, the final report of the subject on the presence or absence of a stimulus. This hypothesis postulates the existence of a decision variable that would accumulate mainly consciousness-related neural activity, but also, to a lesser extent, attention-related neural activity. Behavioral reports based on this decision variable could therefore show an interaction between attention and consciousness, whereas neural variables could be related solely to attention and consciousness.
Intuition favours the gateway model but Tallon-Baudry puts forward evidence for the cumulative model but looking for connections (or lack of connections) in the neural correlates of attention and consciousness. If they have no joint neural correlates then how can they influence one another. The obvious overlaps are neural amplification, involvement of the fronto-parietal network, and oscillatory synchrony; none of these appear solid.
To summarize, attention does operate in sensory regions, but neural amplification by attention appears functionally distinct from the neural amplification related to consciousness: attention does not shorten response latencies, as more contrasted objects would, attention-related and consciousness- related neural activities in retinotopic areas can be dissociated . There is growing evidence that events that do not reach consciousness nevertheless activate parietal and frontal regions, suggesting that they are not sufficient for consciousness to emerge. Because frontal regions are not always activated, one can even wonder whether they are necessary. An alternative possibility is that they reflect a consequence of consciousness, rather than a cause. Last, oscillatory synchrony is not associated exclusively with a single process, be it feature binding, memory and learning, attention, or consciousness, but should rather be considered as a generic mechanism governing neural interactions .
Tallon-Baudry points out a few of the weaknesses of the cummulative model (the place of decisional bias, the dangers of fMRI data, the other types of attention besides spatial) but insists that to the present it best fits the evidence.
Even if the cumulative influence hypothesis has its limitations, it is so far the model that fits best with experimental data. Interestingly this model points toward the existence of a sensory neural activity related to consciousness, uncontaminated by other cognitive processes such as attention. Such an activity could potentially be very close to the immediate subjective experience of the subject. This is reminiscent of the idea of phenomenal awareness, that could be distinct from cognitive access.
V. Wyart, S. Dehaene, & C. Tallon-Baudry (2012). Early dissociation between neural signatures of endogenous spatial attention and perceptual awareness during visual masking Frontiers in Human Neuroscience DOI: 10.3389/fnhum.2012.00016
Catherine Tallon-Baudry (2012). On the neural mechanisms subserving consciousness and attention Frontiers in Psychology DOI: 10.3389/fpsyg.2011.00397