The representation of space in mental models derived from text

Area of Science:

• Cognitive Psychology
• Human-Computer Interaction
• Natural Language Processing

Background:

• Mental models represent text situations, often theorized to function in a Euclidean space.
• Proximity in mental models is hypothesized to enhance short-term accessibility of related information.
• The 'noticing hypothesis' suggests close objects in a mental model are more easily accessed.

Purpose of the Study:

• To test the 'noticing hypothesis' regarding spatial mental models.
• To investigate if object proximity in a mental model aids accessibility of implicitly related information.
• To explore factors influencing spatial representation formation during reading.

Main Methods:

• Participants read texts describing spatial layouts with varying object proximity.
• Seven experiments manipulated factors like text complexity, diagrams, modality, and spatial dimensions.
• Accessibility of target objects was measured under different conditions.

Main Results:

• No consistent support was found for the noticing hypothesis.
• Spatial representations are formed during reading.
• Automatic noticing of implicit spatial relations was not supported.
• The spatial representation appears more topological/functional than Euclidean.

Conclusions:

• Mental models of text do not automatically leverage spatial proximity for information access.
• Reading can create spatial representations, but they lack automatic inference mechanisms.
• The functional or topological nature of these representations warrants further investigation.