Hierarchical resource rationality explains human reading behaviour

Area of Science:

• Cognitive Science
• Neuroscience
• Computational Linguistics

Background:

• Reading is a complex cognitive task requiring coordinated eye movements for comprehension.
• Current theories inadequately explain the link between eye movements and reading comprehension.
• Understanding this link is crucial for cognitive models of reading behavior.

Purpose of the Study:

• To propose a unifying control principle for adaptive reading behavior.
• To explain how eye movements are coordinated to optimize reading comprehension under resource constraints.
• To bridge the gap between theories of eye movements and theories of reading comprehension.

Main Methods:

• Developed a resource-rational optimization framework for reading.
• Created a computational model implementing this framework.
• Validated the model against a wide range of human reading data, from lexical processing to comprehension outcomes.

Main Results:

• The resource-rational model successfully replicates diverse findings in human reading.
• Eye movements are shown to be hierarchically organized to balance comprehension and cognitive costs across different time scales.
• The model accounts for lexical effects, sentence integration, and text-level comprehension strategies.

Conclusions:

• Resource rationality offers a unifying principle for adaptive reading behavior.
• This framework provides a principled account of how cognitive skills adapt to limited resources during knowledge-intensive tasks.
• Suggests a general mechanism for coordinating perception, memory, and action in complex human behaviors.