Theories of access consciousness

Selective attention acts as a filter, retaining only a fraction of the input. Consciousness would be associated with the further processing of these inputs, chosen for their relevance to the organism's goals. The hypothesis of a central limit recurs, in distinct forms, in several theories of the "central bottleneck" (Pashler, 1994), the "two stages of perception" (Chun & Potter, 1995), or perceptual awareness limited to a "Boolean map" (Huang, Treisman, & Pashler, 2007).

It should be noted that these theories do not imply an identity between the processes of attention and awareness. If we stick to the definition of attention as the selection of an object or spatial position, and the amplification of its sensory attributes, a large body of experimental data shows that these processes can unfold without leading to conscious perception. Indeed, alerting and redirecting effects can be induced by subliminal cues (Bressan & Pizzighello, 2008; Woodman & Luck, 2003). Attention can amplify the processing of stimuli that nevertheless remain subliminal (Naccache, Blandin, & Dehaene, 2002). Finally, the cerebral correlates of selective attention and access to awareness of stimuli presented at threshold are radically different (Wyart & Tallon-Baudry, 2008). Thus, attention and awareness are dissociable - but in the presence of multiple stimuli, selective attention appears to be an essential prerequisite, necessary but not sufficient - the "gateway" to conscious processing (Mack & Rock, 1998).

The importance of late and top-down amplification. Gerald Edelman sees "reentry" - the permanent, dynamic, recursive exchange of parallel signals between brain maps - as an indispensable mechanism of unified perception. Certain forms of global reentry in the cortex could lead to consciousness. The idea is developed by Giulio Tononi, according to whom conscious information is both integrated (unified into a whole) and differentiated (the number of contents of consciousness is gigantic, and each of them is highly informative). Differentiation results from the multiplicity of coding neuronal groups, distributed in numerous regions of the cortex, while integration emerges from the many reentrant loops that link them. It can be measured mathematically by the parameter Φ, which evaluates the minimum mutual information between two sub-parts of the complete system. Note that this theory predicts an absence of precise cerebral localization: consciousness would be associated with a distributed thalamo-cortical "dynamic core" whose contours would constantly change and vary considerably from one person to another. Neurophysiologist Victor Lamme, on the other hand, developed a more neurally precise model. He postulates that local recurrent processing, resulting from the conjunction offeedforward sweep andtop-down amplification, enables phenomenal awareness of sensory information. Access to this information is only possible at a later stage, via global recurrent connections to the cortex.

An internal space for synthesizing, maintaining and sharing data. The metaphor of the conscious theater was proposed by Taine as early as 1870: "The mind of a man may be compared to a theater of indefinite depth, whose ramp is very narrow, but whose stage widens from the ramp. In front of this lit ramp, there's hardly room for more than one actor... Beyond, on the various planes of the stage, are other groups all the less distinct for being farther from the ramp." Criticized by Dennett (see Dennett, 1991), this metaphor is nevertheless developed by Bernard Baars (1989) in his model of the conscious workspace. According to Baars, a large number of specialized processors operate non-consciously in parallel, while at any given moment, a single coalition of dominant processors sends its output to the global workspace, where it is broadcast to the entire system. Consciousness thus acts as a space for synthesizing, maintaining and sharing data.