To identify cortical regions associated with neural processing load related to semantic ambiguity using a non-verbal paradigm.

The neural mechanisms of semantic processing are difficult to delineate from lexical, acoustic, and speech-motor neural activity since most relevant neuropsychological tests require language. Furthermore, distinct signal properties of different cortical semantic processing regions may require a more comprehensive analysis of neural signal components than standard metrics such as BOLD/high gamma signals.

We utilized a recently described Visual-based Semantic Association Task (ViSAT, purely image-based), derived from the Pyramids & Palm Trees and Camels & Cactus tests, scaled to 100 unique trials using real-world color pictures to avoid linguistic/articulatory confounds. We crowdsourced normative data to create a metric of trial-by-trial difficulty levels (semantic ambiguity) through Amazon MTurk workers. ICEEG data was recorded from sixteen patients with drug-resistant epilepsy implanted with grid, strip, and depth electrodes predominantly over the lateral cortical regions. Single-electrode neural activity was modeled as a function of semantic ambiguity processing load.

Trial-based semantic ambiguity (percent of responses choosing the top answer, ranging from 54.5–100%) was best approximated by a log(1-x) transformation to approach normality. Electrodes showing increased power with semantic ambiguity (β > 99.9% C.I., cluster-based permutation test) temporally aligned with stimulus and answer (semantic processing) stages were identified, with lower frequency bands (5-20 Hz) showing more robust effects. These cortical sites showed highly variable and transient involvement but generally involved temporal-parietal association areas – particularly the left middle and posterior MTG, angular gyrus, IFG, PHC, and posterior superior frontal cortices.

A non-verbal semantic association task revealed multiple cortical regions encoding semantic difficulty through diverse neural signal properties including low frequency oscillations. Rather than specific local neuronal population firing patterns, these findings suggest that ambiguity in semantic processing may be manifested through network oscillations reverberating throughout distant yet interconnected cortical network hubs.