Please enable JavaScript.
Coggle requires JavaScript to display documents.
ERC GENERAL OUTLINE (conceptual modules (visual feature diagnosticity…
ERC GENERAL OUTLINE
-
vary the mapped semantic dimensions' (words) specificity/diagnosticity (e.g. the sem. space covered by its nearest neighbors). This could allow for finding out if the brain starts of with very specific or very general associations
I might need to make sure to select a set of words such that it is able to dissociate specificity from diagnosticity.
Related: since I'm only interested in concrete nouns I could consider constraining the space to those only to get an optimized fit (and simplify matters...) -> might lead to a lot of information loss...
-
Spatial Question for fMRI How is visual and semantic feature diagnosticity encoded in brain areas along the visual hierarchy (and how do they relate to each other)?
-
-
-
-
-
-
-
-
-
Temporal question for EEG: How does the chronology of semantic and visual features diagnosticity encoding look like and relate to eachother? (e.g. forests before the trees -> pred coding?)
Individual differences question - including ASD To which extent does neural SFD and VFD processing depend on individual differences and AQ sqores (and is there a sharp distinction between clinical and non clinical AQ scores)
Related seminal papers:
Oliva&Schyns (1997) Coarse Blobs or Fine Edges? Evidence That Information Diagnosticity Changes the Perception of Complex Visual Stimuli
Evidence for similar patterns of neural activity elicited by picture- and word-based representations of natural scenes.
Kumar M1, Federmeier KD2, Fei-Fei L3, Beck DM2. (2017)
-
SpatioTemporal question for fMRI-EEG: NEEDS TO BE SPECIFIC AND IMPORTANT TO MOTIVATE METHOD how much information needs to be present in which brain areas to trigger "object recognition"
-
-
n-words (concr nouns) that can maximally discriminate the 100 image labels (in GV space) OR see unsorted tree
-
fMRI responses to each word, probably with a semantically engaging task
-
VFD based on partial-feature set presentations and reverse correlation. Task could be an object naming task (Clarke&Taylor)
With a speeded simple semantic task like animate/inanimage -> this might be important to relate EEG-fMRI to object recognition
-
-
-
For each visual object I could select n (e.g 10) words that are closely associated with the object (or at least closer than all other words) but which vary with regard to their specificity. (controlling for things like word frequency).