English
 
Help Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Poster

How expectation modulations based on verb bias and grammatical structure probabilities shape the sentence processing network

MPS-Authors
/persons/resource/persons138051

Weber,  Kirsten
Hanse Institute for Advanced Studies, Delmenhorst;
Neurobiology of Language Department, MPI for Psycholinguistics, Max Planck Society;

External Resource
No external resources are shared
Fulltext (restricted access)
There are currently no full texts shared for your IP range.
Fulltext (public)
There are no public fulltexts stored in PuRe
Supplementary Material (public)
There is no public supplementary material available
Citation

Weber, K., Micheli, C., Ruigendijk, E., & Rieger, J. (2016). How expectation modulations based on verb bias and grammatical structure probabilities shape the sentence processing network. Poster presented at the Eighth Annual Meeting of the Society for the Neurobiology of Language (SNL 2016), London, UK.


Cite as: https://hdl.handle.net/11858/00-001M-0000-002B-0BDE-4
Abstract
During language processing we use available information to facilitate the processing of incoming information. This allows for efficient processing of predicted information but also causes prediction error when expectations are unfulfilled. In the present experiment we manipulated two types of information in the linguistic input, the statistics of the input (the proportion of two sentence structures across different blocks) as well as verb cues biasing towards one structure or the other. We investigated the brain networks involved in using these two types of contextual information during language processing. We acquired fMR images while 20 participants read ditransitive sentences in their native language (German). The sentences were created with verbs that either biased towards a prepositional object (PO) structure (e.g. “sell” biases towards PO sentences like “The girl sold the book to the boy”) or a double object (DO) dative structure (e.g. “show” biases towards a DO structure like “The girl showed the boy the flower”). The grammatical structure probabilities were manipulated across blocks (Block 1: 75% DO structure, 25% PO; Block 2: 25% DO, 75% PO; block order was counterbalanced across participants). In each block these constructions occurred with equal amounts of DO and PO-biased verbs. After 12% of the sentences a comprehension question was asked to ensure attention. SPM12 was used for fMRI activation and the gPPI toolbox for functional connectivity analyses (flexible-factorial models, voxel-level threshold p<.005, cluster-level pFWE<.05). The anterior cingulate cortex (ACC), a region thought to monitor changes in statistical contingencies, in consort with regions of the language network in temporal and inferior frontal cortex, appears to prepare the network for what is most likely to appear next. The ACC was more tightly coupled to the left posterior temporal cortex as well as the left temporal pole and left inferior frontal gyrus for sentences containing a verb cue towards the overall more likely structure in German (DO). Connectivity to left middle and anterior temporal cortex on the other hand was strongest when the verb cue biased towards the structure that was currently more frequent according to the statistics of the current block. Within the language network, regions of interest related to syntactic processing in left inferior frontal gyrus and left posterior middle temporal gyrus showed greater activation if a prediction based on the verb cue was not fulfilled. This prediction error effect was larger if the verb biased towards the less frequent structure in every day statistics (PO). For the DO verb cue on the other hand, an effect of unfulfilled prediction was seen when the DO was the more infrequent structure in a block and localised to left posterior temporal/occipital cortex and the ACC. In sum, the language network dynamically shifts its connectivity profile in accordance with different information that can be used for building up predictions. Within those regions unfulfilled predictions cause a prediction error, which is modulated by the reliability of the previous prediction. This research was supported by the SFB-TRR31 fund and a fellowship at the Hanse Institute for Advanced Studies to KW.