Evidence for an All-or-None Bifurcation During the Attentional Blink

Several theories of the neural correlates of consciousness assume that there is a continuum of perception, associated with a gradual change in the intensity of brain activation. But some models, considering reverberation of neural activity as necessary for conscious perception, predict a sharp nonlinear transition between unconscious and conscious processing. We asked participants to evaluate the visibility of target words on a continuous scale during the attentional blink, which is known to impede explicit reports. Participants used this continuous scale in an all-or-none fashion: Targets presented during the blink were either identified as well as targets presented outside the blink period or not detected at all. We suggest that a stochastic nonlinear bifurcation in neural activity underlies the all-or-none perception observed during the attentional blink. Whether there can be a strict dissociation between conscious and unconscious processing is a debated issue. Some imaging studies of visual perception show a gradual increase in the cortical activity evoked by a stimulus as participants report increased knowledge of the stimulus (Bar et al., 2001; GrillSpector, Kushnir, Hendler, & Malach, 2000; Moutoussis & Zeki, 2002). For example, Bar et al. (2001), in a study on object recognition, observed a progressive increase in brain activation in several areas of the anterior fusiform gyrus as recognition level increased. According to Farah (2000), ‘‘consciousness may be associated only with the higher-quality end of the continuum of degrees of representation’’ (p. 295). As noted by Kanwisher (2001), signal detection theory and connectionist models have contributed to promote the idea that mental representations are graded rather than discrete. Other studies, however, challenge this view by showing large all-or-none changes in neural activity when a stimulus fails to be reported compared with when it is reported (Dehaene et al., 2001; Lamme, Super, Landman, Roelfsema, & Spekreijse, 2000; Super, Spekreijse, & Lamme, 2001). Indeed, a qualitative difference between unconscious and conscious processing is generally predicted by theories that view recurrent interactions between distant brain areas as a necessary condition for conscious perception (Dehaene, Kerszberg, & Changeux, 1998; Dehaene & Naccache, 2001; Di Lollo, Enns, & Rensink, 2000; Lamme, 2003; Lamme & Roelfsema, 2000). According to one of these theories (Dehaene et al., 1998), consciousness is associated with the interconnection of multiple areas processing a stimulus through a ‘‘neuronal workspace’’ (Baars, 1989, 1997) within which recurrent connections allow long-distance communication and auto-amplification of activation. Neuronal network simulations (Dehaene, Sergent, & Changeux, 2003) suggest the existence of a fluctuating dynamic threshold. If the primary activation evoked by a stimulus exceeds this threshold, reverberation takes place and stimulus information gains access, through the workspace, to a broad range of areas, allowing, among other processes, verbal report, voluntary manipulation, voluntary action, and long-term memorization. If the activation is below this threshold, however, stimulus information remains unavailable to these processes. Thus, this theory predicts an allor-none transition between conscious and unconscious perception (see Fig. 1). More generally, many nonlinear dynamic systems with self-amplification are characterized by the presence of discontinuous transitions in internal state, so-called catastrophes (Thom, 1972). Address correspondence to Claire Sergent, INSERM U 562, Service Hospitalier Frédéric Joliot, CEA/DRM/DSV 4, place du Général Leclerc, 91401 Orsay Cedex, France; e-mail: [email protected]. PSYCHOLOGICAL SCIENCE 720 Volume 15—Number 11 Copyright r 2004 American Psychological Society In the present study, we tested the all-or-none character of conscious perception using an attentional blink (AB; Raymond, Shapiro, & Arnell, 1992) paradigm. The AB is observed when two targets are embedded in a rapid sequence of distractors: Correct identification of the first target (T1) hinders explicit report of the second target (T2) if they are separated by 200 to 500 ms (Broadbent & Broadbent, 1987). The AB affects a vast range of explicit tasks on T2, but the behavioral measures currently used to detect the AB (accuracy on a forced-choice task) do not allow one to determine whether participants are really unconscious of that target, especially given that accuracy is often slightly above chance level. We examined whether the AB merely degrades the available information on T2 or corresponds to an all-or-none loss of conscious perception of T2. To this end, we asked participants to rate the visibility of T2 on a continuous scale. Using a continuous measure instead of discrete response categories (Bar et al., 2001) allowed us to test the continuous or discontinuous character of perceptual transitions (Massaro & Cohen, 1983). To rule out the possibility that information on T2 had been forgotten by the time the question was asked, we required that the subjective response be made immediately after the presentation of T2. EXPERIMENT 1: ALL-OR-NONE RESPONDING IN AN AB PARADIGM We first studied the use of the continuous scale in a classical AB paradigm in which the stimulus-onset asynchrony (SOA), or lag, between T1 and T2 was varied. If the AB merely degrades the available information on T2, the distribution of participants’ responses would be expected to shift gradually from the low end of the visibility scale when the AB was strongest (SOA of around 300 ms, corresponding to lag 3 in our experiment) to a higher level of visibility as lag increased. However, if the AB reflects increased probability of an all-or-none loss of conscious perception, responses would be expected to distinguish two types of trials: not-seen trials, on which T2 completely failed to be consciously perceived, and seen trials, on which T2 was fully perceived. Thus, the distribution of perception ratings would be bimodal. Furthermore, T2 visibility in the seen trials would be unaffected by the T1-T2 lag. Instead, the lag would affect the relative proportions of seen and not-seen trials.