Graded expectations: Predictive processing and the adjustment of expectations during spoken language comprehension

Area of Science:

• Cognitive Neuroscience
• Psycholinguistics
• Computational Linguistics

Background:

• Listening comprehension relies on integrating incoming words with existing contextual information.
• Both local (word-level) and global (scenario-level) contexts influence word processing.

Purpose of the Study:

• To investigate how local and global contextual cues are utilized during listening comprehension.
• To examine the neural correlates (event-related potentials) of context integration.

Main Methods:

• Manipulating local context with prototypical/atypical feature words preceding a target noun.
• Manipulating global context by embedding the target noun in consistent/inconsistent scenarios.
• Analyzing event-related potentials (ERPs), specifically N400 and N200 components, at critical time points.

Main Results:

• An N400 effect at the feature word indicated global context compatibility.
• An interaction between global predictability and local consistency modulated the N200 and N400 at the target noun.
• Post-N400 positivity reflected context updating when only one contextual cue was supportive.

Conclusions:

• Incoming words compatible with context-based expectations are processed more efficiently.
• Specific lexical items can be activated when context is highly constraining.
• Listeners dynamically adjust expectations based on contextual support, even with inconsistencies.