Dissociating 'what' and 'how' in visual form agnosia: a computational investigation

Area of Science:

• Neuroscience
• Cognitive Psychology
• Computational Neuroscience

Background:

• Visual form agnosia is characterized by impaired shape perception (recognizing 'what' an object is) despite intact visuomotor abilities (acting 'how' to use an object).
• This 'what-how' dissociation presents a challenge in understanding how patients can interact with objects they cannot consciously perceive.
• Existing explanations for this dissociation have limitations.

Purpose of the Study:

• To present and test a novel 'pathway information' account for the 'what-how' dissociation in visual form agnosia.
• To investigate if task information load explains preserved action abilities despite impaired perception.

Main Methods:

• A neural network model was developed to simulate visual processing pathways.
• The model sustained damage to a common input layer shared by 'what' and 'where/how' processing pathways.
• Model performance was evaluated on object identification ('what') versus spatial localization ('where/how') tasks.

Main Results:

• Simulated neurological damage disproportionately impaired object identification compared to spatial localization.
• The model's performance mirrored the 'what-how' dissociation observed in patients with visual form agnosia.
• Results support the hypothesis that 'what' tasks require more information than 'where/how' tasks.

Conclusions:

• The 'pathway information' account provides a parsimonious explanation for the 'what-how' dissociation in visual form agnosia.
• Differential information requirements across visual pathways may explain spared visuomotor function in agnosia.
• Findings have implications for understanding visual form agnosia and related neuropsychological syndromes.