You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

 
Poster C42, Thursday, November 9, 10:00 – 11:15 am, Harborview and Loch Raven Ballrooms

Extracting Single Word Voxel Patterns from Self-Paced Reading using Simultaneous Eye-Tracking and Multiband fMRI

Benjamin Schloss1, Chun-Ting Hsu1, Ping Li1;1Pennsylvania State University

When stimuli are highly similar in either sensory modality or word category, it becomes increasingly difficulty to differentiate elicited brain responses based on the activity of a particular region. Therefore, patterns of activity provide researchers with a more sensitive way to assess how physiological activity in the brain changes in response to more subtle changes in the external environment. For example, reading pseudowords may activate entirely different brain regions than reading real words (Dehaene et al., 2010). However, reading about different types of animals may only elicit slightly different patterns of information within the same regions. It has also been shown that pattern information changes as a function of learning (Bauer & Just, 2015), such that new information systematically changes the functional patterns elicited by a word over a matter of minutes during reading. In sum, the current research suggests that the adult brain is capable of updating complex and distributed information maps in an almost continuous fashion. However, relatively little is known about the online mechanisms that support this continuous information processing, especially in higher level, human specific cognition such as language. Traditionally, questions about fast online processing mechanisms and specific anatomical locations of certain brain functions have been pursued by independent lines of research due to methodological constraints (e.g., using eye-tracking or MEG or EEG to assess the temporal dynamics of reading, or fMRI to identify the spatial representation of meanings). In the current study, we asked whether it is possible to estimate spatially detailed patterns of brain activity in response to a single word during self-paced reading. On average, skilled readers fixate on a word for only 200-250 ms and semantic information is thought to be accessed around 400 ms after processing of a particular word begins (Rayner, 1998). Traditionally speaking, this level of speed falls outside of the temporal resolution of most fMRI experiments. Using representational similarity analysis techniques (Kriegeskorte et al., 2008), we provide visual and statistical evidence of methodological advancements which allow researchers to estimate the pattern of activity elicited by a single word during naturalistic reading across the entire cortex brain using a 3x3x4 mm voxel size. The current study had 50 participants. Using simultaneous eye-tracking and multiband fMRI, we achieve a 400 ms TR, and are able to accurately estimate the onset of individual word processing without controlling the timing of the experiment. In particular, we replicate a previous study (Anderson et al., 2015) which used slow event related design to estimate patterns of activity for concrete concepts, showing visual and linguistic organization of concepts along a number of regions in the what pathway as well as statistically reliable difference in the amount of visual and linguistic information present in respective regions in the visual and language networks. The current relevance of the methodological advancement is discussed in terms of its potential to inform theories of semantic learning as well as its potential for translating basic neurobiology of language research to more complex and realistic environments.

Topic Area: Meaning: Combinatorial Semantics

Back to Poster Schedule