Abstract
This research investigates the cognitive "situation models" readers construct during text comprehension. By manipulating the spatial relationship between a protagonist and a target object, the study demonstrates that objects physically associated with a character remain in the mental foreground. Consequently, processing information that contradicts these active mental models incurs a significant cognitive cost, manifesting as slower reading and reaction times. The findings suggest that human text processing is a dynamic, structural process rather than a simple analysis of linguistic propositions.
Introduction
When we read, do we merely store a list of facts, or do we build a "mental movie" of the scene? Cognitive psychologists refer to these mental representations as situation models—analog, spatial maps of the narrative. A central question in psycholinguistics is how these models influence our ability to handle inconsistent information. This paper explores whether the difficulty of processing a contradiction depends on whether an object is currently "in focus" (e.g., being worn by a character) or "out of focus" (e.g., left behind in another room).
Methods
The researchers conducted a controlled experiment with 32 participants using a series of short, five-sentence stories. Each story followed a rigorous template designed to manipulate the reader's mental model:
- Setting: Introduces the protagonist and a target object (e.g., a sweatshirt).
- Manipulation (SM-Relation): The object is either associated with the protagonist (e.g., put on) or dissociated (e.g., taken off).
- Fillers: Sentences describing the protagonist moving to a new location, keeping the focus on the character while the object is not mentioned.
- Critical Sentence: A statement that is either consistent or inconsistent with the object's current state.
- Verification: A True/False task regarding the object's location to measure reaction time and accuracy.
The study utilized 8 critical stories and 16 distractors, employing a Latin Square design to ensure participants encountered all experimental conditions without repeating specific narratives.
Key Insights and Ideas
- The Foregrounding Effect: Objects physically linked to a protagonist (like a bag being carried) are "foregrounded" in the situation model. They effectively "travel" with the protagonist in the reader's mind, remaining highly accessible.
- The Cost of Inconsistency: Processing inconsistent information is a structural clash within the mental model. When new text contradicts the established scene, reading speed drops as the brain attempts to resolve the conflict.
- Mental Search vs. Immediate Clash: In the "dissociated" condition, objects eventually drop out of the active situation model. If a later sentence provides inconsistent info about a "forgotten" object, the brain must perform a costly "memory search" to retrieve the original context, significantly slowing reaction times.
- Strength of Association: The effects are most pronounced when the spatial link is "strong" (e.g., direct physical contact) compared to "weak" associations (e.g., an object hidden in a pocket).
Key Results
- Reading Times: There was a significant main effect of consistency ($p < .05$). Consistent sentences were read faster than inconsistent ones, indicating that the brain immediately detects model-level contradictions.
- Reaction Times: A significant interaction effect ($p < .05$) was found in the verification task. In the "dissociated" condition, inconsistent information led to the slowest reaction times, averaging ~5000ms compared to ~3400ms for consistent information.
- Strong Spatial Links: Sub-analysis showed that reading times were longest for the "associated-inconsistent" condition, where the clash with the active, foregrounded mental model was most direct.
Conclusion and Future Work
The study confirms that readers build dynamic situation models that prioritize spatially associated objects. However, the authors noted significant heterogeneity between different stories, suggesting that subtle nuances in description can drastically alter model construction. Future research should focus on stricter linguistic balancing and explore how these spatial manipulations affect long-term retention versus immediate processing.
Summary updated