PhD Project 6
Sharpening sensory predictions by linguistic primes
(last update 2019-06-27)
Studies within the visual domain have shown that top-down prediction facilitates perception by sharpening the response to expected percepts in early visual cortex. Here, the aim is to investigate whether this also holds for cross-modal priming between language and vision. Words will be used to prime visual stimuli and measure magnitude and sharpness of responses in primary and high-order visual areas. It will be studied how the prediction for specific visual features, e.g. face gender characteristics, depends on vocabulary. Predictions will be manipulated by varying the frequency of congruent prime-target pairs, and by using primes with different levels of specificity.
Team members: Todorova, Neville, and Piai
This project is dedicated to the investigation of executive control in language-based decision-making. It was shown that disentangling semantic from executive processes has important implications for language theory. The topic was investigated using neuroimaging techniques (fMRI, MEG) and various quantitative methods (model based and data driven methods).
When asked to match a word to a picture, participants are slower and less accurate for related word-picture pairs (word: banana, picture: apple) relative to unrelated pairs (word: banjo, picture: apple). The underlying nature - semantic or executive - of this interference effect is still debated. Analysis was performed on the word-picture matching (WPM) performance of patients with stroke-induced lesions to the left-temporal or left-frontostriatal cortex and matched controls using diffusion drift model (DDM). WPM performance in controls tapped into the amount of information (decision boundary) needed to make a decision: A higher threshold was associated with related relative to unrelated word-picture pairs, but there was no difference in the quality of the evidence (drift rate). This suggests an executive rather than semantic mechanism underlying WPM interference. Patients with lesions to left-temporal cortex showed no differences either in drift rate or decision boundary. By contrast, patients with frontostriatal lesions exhibited both increased drift rate and decision boundary for unrelated relative to related pairs. Left-frontostriatal and temporal damage affects the computations required by WPM differentially, resulting in systematic deficits across lexical-semantic memory and executive functions. These results support a neuronal dissociation between lexical-semantic memory and semantic control mechanisms.
Left panel. Posterior estimates of the hierarchical drift-diffusion model for the drift rate parameter. The difference in the related vs. unrelated condition was not detected in control participants but in patients with both temporal and frontal lesions.
Right panel. Posterior estimates of the hierarchical drift-diffusion model for the decision threshold parameter. Related vs. unrelated differences in decision threshold had the opposite directionality for control participants as opposed to participants with temporal or frontal lesions.
Whole-brain dual-coded t-maps that simultaneously show the contrast estimate and the un-thresholded t–statistic (p<.001 UNC), to observe relative BOLD signal for +/- interaction effects of interest. [+ red] [- blue] (Plots taken from fMRI study 2)
The project consists of four experiments that have been successfully completed. The scientific papers highlighting the results of the project have been submitted to peer-reviewed journals or are in the preparatory phase. It was shown that disentangling semantic from executive process has important implications for language theory.
The project team incorporates a unique combination of expertise that represents team science.
Cutting-edge scientific research in the language field combines with mathematical formalism that has been traditionally used in the field of perception and decision making. This project brings together experts from different fields using various experimental methodologies, which allows studying complex problems at various level of description.