The structure of integral dimensions: contrasting topological and Cartesian representations

Area of Science:

• Cognitive Psychology
• Perception Science
• Computational Neuroscience

Background:

• Perceptually integral dimensions (e.g., color) are typically represented holistically.
• Existing models (multidimensional scaling, general recognition theory) use Cartesian coordinates for both integral and separable dimensions.
• The psychological validity of Cartesian models for integral dimensions remains unclear.

Purpose of the Study:

• To investigate the underlying structure of holistic representations for integral perceptual dimensions.
• To compare the predictive power of Cartesian versus topological models for integral dimensions.
• To examine how perceptual learning (dimension differentiation) shapes these representations.

Main Methods:

• Experimental testing of Cartesian and topological models.
• Utilizing the perceptual-learning phenomenon of dimension differentiation.
• Designing discrimination training tasks with integral-dimension stimuli.

Main Results:

• Results consistently supported the Cartesian model over the topological model.
• Learned analytic representations of integral dimensions aligned with Cartesian predictions.
• Dimension differentiation training induced structured, rather than purely holistic, representations.

Conclusions:

• Perceptual representations of integral dimensions possess a surprisingly structured nature.
• Cartesian coordinate systems provide a more psychologically meaningful framework than topological spaces for these dimensions.
• Perceptual learning can transform holistic representations into more analytic, structured forms.