Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams
Modality Independent Representations of Conceptual Categories in Picture Naming and Word Reading
Poster A17 in Poster Session A, Thursday, October 6, 10:15 am - 12:00 pm EDT, Millennium Hall
Julien Dirani1, Liina Pylkkänen1,2; 1New York University, 2NYU Abu Dhabi
In speech production, meanings are transformed into sounds. For example, when naming a picture of a dog or reading the word “dog” out loud, a speaker typically activates a conceptual representation about dogs. While a picture of a dog illustrates a specific exemplar, the word “dog” refers to the class of dogs. It remains an open question whether shared representations are activated prior to speech independently of the input modality or whether those representations are modality specific. In fact, while some theories hold that concepts are exclusively encoded via the perceptual system (Barsalou, 1999), others suggest the existence of amodal representations (Mahon & Caramazza, 2008; Ralph, Jefferies, Patterson, & Rogers, 2017). Here we addressed this question by asking 24 participants to name pictures and read words aloud during an MEG measurement and by decoding for a tool vs. animal distinction in the MEG data with generalization both across time and stimulus modality (King & Dehaene, 2014). Crucially, the visual forms of the words did not contain information about their category-membership and the language production task did not explicitly require a categorization judgment. This allowed us to investigate whether and when modality-independent conceptual representations spontaneously come online in each of the two language production tasks. We found that while evidence of modality-specific category representations emerges at 75ms post-stimulus onset for picture naming and at 95ms for word reading, modality independent representations are active slightly later at 125ms for pictures and at around 150ms for words. Further, modality-specific representations evolved as a feedforward process for both modalities, while modality independent representations were active simultaneously early in the timecourse, while later, they were delayed in picture naming compared to word reading. A follow up analysis indicated that our decoding of semantic categories is very unlikely to be confounded with information related to the word forms. Finally, we explored the spatial and temporal evolution of category information in the MEG sensors for each of the overt reading and picture naming tasks. All together, these findings provide evidence for the spontaneous activation of modality-independent representations of categories in picture naming and word reading, thus supporting theories of concepts in which modality independent representations exist (Mahon & Caramazza, 2008; Ralph, Jefferies, Patterson, & Rogers, 2017).
Topic Areas: Methods, Language Development/Acquisition