Conference Agenda

The common approach to studying unconscious processing has recently been criticized for a fundamental statistical fallacy: researchers conclude unconscious processing when a measure of stimulus awareness (e.g., discrimination ability) does not exceed chance performance while a measure of stimulus processing (e.g., priming effects or EEG responses) is significant – without directly comparing the two. Most landmark findings of unconscious processing fail to meet the critical statistical criterion of stimulus processing exceeding awareness.

In a series of priming and EEG experiments involving 738 subjects, we determined whether metacontrast-masked shapes can be processed unconsciously adopting this direct-comparison criterion. We also addressed other pressing issues in the scientific study of unconscious processes, including unequal statistical power of the two measures, post-hoc subject selection, and motivational factors. Independent data sets were used to identify and test a subset of “unconscious” participants for whom metacontrast masking rendered the shapes indiscriminable.

For this unconscious subsample, processing and awareness measures were transformed to the same scale reflecting each measure's sensitivity. Priming effects, measured via speeded go/no-go tasks, consistently exceeded prime discriminability across all behavioral experiments. Similarly, multivariate EEG decoding of shapes (~140 ms post-stimulus onset) surpassed behavioral prime discriminability.

These findings provide robust evidence for unconscious processing of shapes and introduce a novel methodological framework that can be used across a wide range of domains to establish a rigorous empirical foundation for future consciousness research.

Unconscious perceptual processing and the extent to which visual working memory operates on unconscious input remain debated, with no definitive evidence confirming or refuting their existence. A key challenge in their empirical investigation lies in methodological constraints, such as criterion problems in reporting (un)awareness. To address these issues, we conducted two studies combining a bias-free two-interval forced choice(2IFC) task with an ensemble perception task that mitigated the criterion content fallacy. We briefly presented ensembles of happy and sad faces, and assessed both perceptual processing and working memory resilience using a delayed response task. We operationalised visual (un)awareness using an objective detection measure from signal detection theory. Results showed a dissociation between awareness and perceptual sensitivity: while participants could not explicitly detect task-relevant ensemble features, they could still discriminate them above chance-level, even after a delay and with distractors, supporting unconscious perception and unconscious working memory processing. To address the potential effects of detection inefficiencies in 2IFC tasks, we conducted a follow-up study using a modified single-interval task with an added distractor interval to disrupt working memory. Considering possible biases in one-interval tasks, we implemented additional criterion checks to validate our findings. These results corroborated our previous findings demonstrating that ensemble perception and working memory can operate at some level in the absence of conscious awareness. Moreover, we observed a similar dissociation between perception and awareness in living and nonliving ensembles, highlighting its generalisability. Ensemble paradigms thus offer a valuable tool for probing the limits of awareness in perception and memory.

Unconscious priming is a highly relevant finding in the field of unconscious cognition. In a classic study by Vorberg (2003), observers were slower to identify the direction of an arrow mask when it followed an arrow prime pointing in the opposite direction (priming effect), despite them not being able to identify the direction of the arrow prime (prime awareness). Although many studies have claimed effects of unconscious priming, the statistical procedure for substantiating such claims is often lacking. For example, the absence of prime awareness is usually established by erroneously accepting the null hypothesis based on a non-significant p-value. Moreover, prime awareness and priming effects are often measured using different scales and not directly compared.

To overcome such problems, we employed a state-of-the-art Bayesian methodological-statistical framework to test priming effects and prime awareness at the single-subject level over multiple experimental sessions. Additionally, participants underwent extensive training in the prime awareness task to ensure they understood and executed the task as intended. Further, we directly statistically compared prime awareness and the priming effect.

By using this approach, we show unconscious processing at short prime-mask SOAs in six out of ten participants, effectively replicating previous results by Vorberg et al. (2003). However, we also show that – unlike in the original study - prime awareness was modulated by the prime-mask SOA, plausibly due to better instruction and training. Altogether, our work replicates a key finding for consciousness research and validates a novel framework to assess unconscious processing.

Although in recent years some studies have found evidence suggesting that working memory (WM) may operate on unconscious perceptual contents, decisive demonstrations of the existence of unconscious WM are lacking. Almost three years ago, we launched a large-scale project to replicate the original study of unconscious WM by Soto et al. (2011). Our Registered Report protocol received in-principle acceptance in Neuroscience of Consciousness in 2024. Data collection is now completed, comprising a multisite sample from 19 laboratories and 40 collaborators, with a much larger number of participants and trials than those used in previous studies (530 participants, with 720 trials each, in two sessions). As a result, a large open-access dataset will be available for researchers and future analyses.

We have critically tested the hypotheses commonly proposed: above-chance performance in cue-present trials reported as unseen, a null correlation between performance and cue detection sensitivity, and a significantly above-chance intercept in the regression of performance on sensitivity. All three results are threatened by measurement error, and consequently, we have calculated reliability estimates for our measures in a series of exploratory analyses, allowing us to diagnose and – if needed – to correct any bias. Our results largely reproduced those of Soto et al. (2011). After three years of work on this international project, the details of these results and conclusions will be available for their first exclusive presentation at this ASSC conference.

During conscious processing of visual scenes, the eyes are consistently attracted to meaningful and visually salient areas. Does this occur also when scenes are rendered invisible? The answer to this question is unknown, as free viewing of invisible scenes has yet to be documented. Here, in two preregistered experiments, fifty-two participants freely viewed scenes presented consciously or unconsciously using Continuous Flash Suppression (CFS) while their eye movements were tracked. Despite being unaware of the scenes, participants gazed more, but not longer, on objects, and were attracted to areas that were both semantically and visually salient. Notably however, their gaze was not attracted to areas that were either visually (but not semantically) salient or semantically (but not visually) salient. A Comparison of gaze patterns to a Convolutional Neural Network (CNN) mimicking the human visual ventral pathway revealed that unconscious gaze patterns correlated only with layers corresponding with early visual processes, while conscious patterns also correlated with higher-level layers. Our results provide first evidence for meaningful gazing on invisible rich and complex scenes, while highlighting the limitations of such processing.