Search Abstracts | Symposia | Slide Sessions | Poster Sessions
How conscious access and task demands shape the neural geometry of word representations during the Attentional Blink
Poster Session B, Wednesday, September 30, 4:30 - 6:30 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Clara Dargent1,2, Rodrigue Reibel3, Wiktoria Januszek1, Nathan Béraud4, Claire Sergent1,2; 1Université Paris Cité, INCC UMR 8002, 75006 Paris, France, 2CNRS, INCC UMR 8002, 75006 Paris, France, 3École Polytechnique, Institut Polytechnique de Paris, 91128 Palaiseau Cedex, France, 4DTIS, ONERA, 13330 Salon-de-Provence, France
The Attentional Blink (AB) paradigm is a powerful tool to make words, presented at supra-threshold durations, "disappear" from conscious access. When performing a task on a first target (T1), the conscious report of a second target word (T2) shown 200 to 500 milliseconds later is indeed severely impaired. Debate persists regarding whether semantic processing is disrupted during this attentional blink window. While classic late-selection accounts suggest that semantic integration remains intact, as evidenced by behavioural priming and the presence of the semantic N400 component, recent studies suggest that semantic extraction might be weakened or altered. Additionally, the neural signatures of conscious language processing remain confounded by task-related processes in traditional designs, since they rely on subjective trial-by-trial reports to assess awareness. The goal of the present study is thus two-fold: (1) explore how conscious access shapes the neural representations of written French words across orthographic, lexical, and semantic levels; and (2) characterize the impact of explicit task demands on these conscious and pre-conscious linguistic representations. We employed a 2×2 within-subject design crossing task demands (Active vs. Passive) and visual awareness (Conscious vs. Non-conscious) during magnetoencephalography (MEG) recording. To isolate task-free conscious and unconscious processing of words, we used a novel passive AB protocol requiring no explicit trial-by-trial report on T2. Instead, conscious access was monitored via intermittent mind-wandering probes and a subsequent incidental memory test, which required participants to distinguish between previously presented and novel words. Here, we will present the behavioural data confirming a robust attentional blink effect induced by T1 in the absence of a task on T2, while also demonstrating how intrinsic lexico-semantic characteristics modulate the probability for a word to access consciousness. Furthermore, we will present preliminary MEG analyses tracking how late evoked components, beyond initial sensory processing phases, are modulated by both access consciousness and task demands. Using Representational Similarity Analysis (RSA) and Multivariate Pattern Analysis (MVPA), we explore word encoding over time across three distinct linguistic levels: orthographic (word length, bigram frequency), lexical (surface frequency in a French corpus, part of speech), and semantic (coarse animacy and more fine-grained semantic categories). This framework will demonstrate whether low-level visual and orthographic dynamics remain stable across conditions, and whether higher-level semantic decoding generalizes. Ultimately, these results will clarify whether conscious access and top-down task demands simply act as a filter within conceptual spaces or reshape the intrinsic geometry of word meaning in the human brain.
Topic Areas: Meaning: Lexical Semantics,