Conference Agenda

A dynamic stimulus presented to one eye can suppress a static target in the other for long periods (continuous flash suppression: CFS). The suppressed target eventually becomes visible and this duration (bCFS) is often used to index unconscious processing. Controversially, faster breakthroughs are considered evidence of visual processing without awareness while opponents claim breakthrough times simply vary with low-level stimulus properties. bCFS times alone cannot solve this: suppression thresholds are needed as a baseline to compare with breakthrough thresholds. Our new ‘CFS tracking’ paradigm (tCFS) quickly measures contrast thresholds for breakthrough and suppression so that suppression strength can be calculated for any image. Participants simply track their changing perceptual states as a suppressed image steadily increases in contrast until visibility (i.e., breakthrough) and then decreases until re-suppression is reported, then increases again (and so on, in a continuing cycle). Using tCFS we confirm that: (i) there are some differences in breakthrough thresholds across target types (e.g., grating vs face), as bCFS has shown, but (ii) suppression thresholds show a parallel pattern of differences, thus (iii) suppression strength is the same for all images (~14-15 dB for gratings, noise, objects, food, faces, biological motion). Uniform CFS suppression strength indicates a single mechanism of CFS suppression, likely early in visual cortex where left and right eyes combine, and prior to processing of objects and image identity. Using the tracking paradigm with binocular rivalry reveals half as much suppression: ~7-8 dB.

Prior studies have shown that (a) a target’s colour information and form information, can be processed without awareness and (b) that unconscious colour processing occurs at early levels in the cortical information processing hierarchy.

Recently, a narrative system has been proposed as an extension to models of consciousness (Lau, 2021). However, it has also been proposed but not tested, that the coherence-seeking narrative system only takes into account what one is conscious of.

In a novel task design, we are testing whether we can dissociate a perceptual choice (immediate response given to the presentation of 3 different visual cues: semantic prime, colour and shape prime, from the integration of exactly the same visual cues into the narrative system (choosing 1 out of 4 available narratives that features none to all visual cues).

In the task, information is presented just below and just above the conscious threshold, so that subliminal and supraliminal trials can be dissociated and the integration of all visual cues into perceptual choice and narrative system can be tested. Thus, we can dissociate information integration into perception and narrative system (justifying the perceptual choice).

We show that there is clear separation between information being integrated into a perceptual choice, but not into the narrative, when presented below threshold for normal observers; while the reverse has been shown for psychotic observers, where more subliminally presented information made it into the narratives, however, there was less coherence / more mismatches between perception and the narratives themselves.

According to the Partial Awareness Hypothesis (Kouider & Dupoux, 2004), some low-level features of presumably subliminal stimuli (e.g., letter shapes) may be consciously perceived by participants in experiments investigating unconscious perception. Such partial awareness may be insufficient for participants to report stimulus visibility, yet it could still activate semantic networks involved in processing of adequate words (or other presented stimuli). As a result, conscious partial perception is a potential explanation for the effects reported in unconscious perception experiments

In this study, we conducted Dehaene et al.'s (1998) experiment, in which reactions to target numbers were influenced by masked numbers. The experiment included two conditions: (1) the exact replication (REP condition) and (2) the eliminated partial awareness condition (EPA condition). In the REP condition, all experimental settings were identical to those in the original study. In the EPA condition, the presentation duration of masked numbers was calibrated individually for each participant in such a way that they had no partial awareness of the stimuli.

We tested 52 participants. In the REP condition, 70% of participants exhibited partial awareness of the masked stimuli. On average, the calibrated duration of the masked stimuli in the EPA condition was 17 ms shorter than in the REP condition. We found no priming effects in the EPA condition, whereas the original priming effect was replicated in the REP condition. These results suggest that partial awareness was responsible for the effects observed in the original experiment, raising questions about the interpretation of studies on unconscious perception.

Visual categorisation of faces, bodies, or scenes relies on partially segregated cortical networks in humans and monkeys. Although the apex of this process has traditionally been ascribed to downstream areas, such as the superior temporal sulcus (STS) and infero-temporal (IT) regions, emerging evidence suggests that early visual areas also provide foundational input for category selectivity via recursive processing. For example, distinct neural subpopulations for faces and bodies are interspaced in occipital areas as early as the primary visual cortex (V1). While category selectivity has been demonstrated in the absence of V1 in humans, few comparative analyses exist in non-human primates. It thus remains unknown how V1 damage modulates both overall neural activity and category-specific responses.

To address this gap, we longitudinally investigated percent-signal change (PSC) across stimulus categories in two rhesus macaques using a contrast-enhanced fMRI protocol before and after a unilateral V1 lesion. We employed a 4 (categories) × 3 (locations) factorial design, presenting central images (15° diameter) of monkey faces, bodies, objects, and their phase-scrambled counterparts at three visual field locations (central and peripheral in the lower left and right quadrants at 17° eccentricity).

Our results reveal that a common neural circuit remains active both pre- and post-lesion. Notably, STS-IT regions that normally exhibit robust category-specific responses lose their selectivity post-lesion. When V1 is compromised, some downstream regions maintain significant activity—albeit with reduced intensity and specificity— but lose their category specificity. These findings highlight the importance of local specialisation and long-range recursive interactions in preserving category discriminability.

Unconscious rewards have been reported to drive behavior, which implies that motivation for action could be processed outside of awareness. Here, we focus on a seminal study by Pessiglione and colleagues (2007) showing that subliminal monetary rewards influence the amount of effort that participants invest in a task. Masked pictures of one pound and one penny coins were shown for variable durations. Participants were rewarded a proportion of the presented coin matching the amount of force they exerted on a handgrip. In a follow-up task, participants reported the identity and visibility of the stimuli. Stimuli lasting up to 50ms were reported to be processed unconsciously. Nonetheless, across all stimulus durations, participants could successfully exert more force when higher rewards were at stake. Similar effects were reported for galvanic skin responses and BOLD responses linked to reward processing.

Given the implications of these results for theories of consciousness and motivation, we set out to replicate it - first by totally adhering to the original protocol (except for the replacement of fMRI with electroencephalography), and, in a successive step, by implementing state-of-the-art methods. At the conference, we will present results from the first study. We could not find behavioral evidence that coin stimuli could be perceived outside awareness, nor any effect of reward magnitude on the amount of applied force (and on all other physiological measures) when subjects were unaware of the rewards. These results call for a deeper inquiry of claims of unconscious reward and motivation processing.

Breaking and Reverse Continuous Flash Suppression represent valuable tools to explore the dynamics of access (bCFS) and disappearance (revCFS) from visual awareness of visual stimuli. In this study, we investigated electrophysiological dynamics of access and disappearance of face stimuli in visual awareness through a time-dependent eXplainable-MultiVariatePatternAnalysis, comparing the EEG activity at each time against a baseline (i.e., the start of the trial) to quantify differences between the baseline and the time-dependent activity. Then, through an explainable AI technique, we defined the relevance of each electrode regarding the baseline-versus-time classification.

In a pilot of 18 subjects, we measured EEG activity during bCFS and revCFS, in which participants provided responses for conscious access and disappearance, respectively. Distinct neural patterns were defined by comparing the dynamics of access and disappearance from awareness within the two tasks. In the first 1000ms, where the face is ramping up from 0 to maximum intensity (suppressed in bCFS, dominant in revCFS), we observed an increased parieto-occipital activity in both bCFS and revCFS, but with greater intensity and earlier in time in bCFS. Comparing the moment of access and disappearance, a sustained parieto-occipital activity, peaking around 400 ms before the report, characterized the access to awareness, whereas conscious disappearance in revCFS was characterized by significant frontal brain activity.

Our results deepen the electrophysiological dynamics of access and disappearance from awareness, showing that conscious access seems mostly modulated by early visual areas. The disappearance from awareness seems rather regulated by frontal areas, with a minor involvement of parietal-occipital regions.