Representations of geometric shapes have syntactic structure

Area of Science:

• Cognitive Psychology
• Linguistics
• Neuroscience

Background:

• Humans may use language-like mental representations for various stimuli.
• Previous evidence relied on indirect measures like complexity and compression.
• Directly probing the structure of geometric shape representations is needed.

Purpose of the Study:

• To directly investigate the structural properties of human geometric shape representations.
• To determine if geometric shapes exhibit linguistic properties like structural ambiguity and constituency.
• To test the hypothesis of hierarchical tree structures in shape perception.

Main Methods:

• Utilized constituency tests adapted from linguistics.
• Conducted three preregistered experiments with controlled geometric shape stimuli (n=136).
• Examined effects of animation, subpart detection, and fragment reconfiguration on shape representation.

Main Results:

• Found robust evidence for hierarchical tree structures in adult shape representations.
• Demonstrated that shape representations vary based on preceding animation.
• Showed that subpart detectability and fragment reconfigurability depend on tree structure.

Conclusions:

• Geometric shapes are internally represented as hierarchical trees of constituents.
• Human shape perception utilizes a language-like representational format.
• Findings support a deeper connection between language processing and visual cognition.