Year: 2026
Type: preprint
Abstract: Reading is sensitive to contextual demands, with readers adapting strategies according to overarching goals and situational constraints. For example, when reading to study or answer questions, readers attend more to goal-relevant sections, read more slowly, and regress more often. While behavioural outcomes are established, the neurocognitive mechanisms enabling this flexibility remain poorly understood. One possible mechanism is the dynamic re-weighting of goal-relevant vs. irrelevant features. To test this, we used eye-tracking and 64-channel EEG to examine how short-term goals modulate online linguistic feature tracking during naturalistic reading. 40 native English speakers read stories in 3 conditions: Baseline (no goal), Orthographic (proofreading goal), and Semantic (comprehension goal). Fixation-aligned EEG was modelled using mTRF to estimate the strength of neural tracking of 3 features: orthographic neighbourhood size (ONS: no. of words differing from target by 1 letter), lexical frequency (Zipf), and Surprisal (word unexpectedness in context) to capture orthographic, lexical, and contextual-level processing. We show that goals selectively modulated all features relative to Baseline: ONS tracking was enhanced in Orthographic condition, Zipf tracking was enhanced in Semantic condition, and Surprisal tracking was inhibited in Orthographic condition. Thus, tracking of orthographic (ONS) and lexical (Zipf) cues were selectively up-regulated when relevant for current goal, while contextual (Surprisal) tracking was selectively down-weighted when task-irrelevant (proofreading). On the mechanistic level, this is consistent with precision weighting accounts of predictive processing. This the first demonstration of top-down goals dynamically re-weighting neural processing of linguistic features, supporting a flexible predictive processing account of goal-directed reading.
Language: English
Subfield: Developmental and Educational Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7233030828
