Neural Tuning for Ordinal Processing: Convergent Patterns in Human Brains and Artificial Networks

Area of Science:

• Neuroscience
• Cognitive Science
• Comparative Psychology

Background:

• Ordinality, the perception of item rank in a sequence, is a fundamental cognitive skill shared across species.
• This ability is vital for behaviors like decision-making, foraging, and social organization, independent of symbolic systems.
• Previous research suggests a neural basis for quantity processing, but the mechanisms for ordinality perception remain less understood.

Purpose of the Study:

• To investigate the neural mechanisms supporting non-symbolic ordinality perception in the human brain.
• To test the hypothesis that neuronal tuning, with neurons selectively responding to specific ranks, underlies ordinality processing.
• To explore whether similar ordinality processing mechanisms exist in artificial neural networks.

Main Methods:

• Utilized ultra-high field 7 Tesla functional magnetic resonance imaging (fMRI) in human participants.
• Applied population receptive field (pRF) modeling to identify neural populations tuned to ordinal positions.
• Trained hierarchical convolutional neural networks on visual tasks to observe spontaneous emergence of ordinality tuning.

Main Results:

• Identified neural populations in parietal and premotor cortices tuned to non-symbolic ordinal positions.
• Observed increased tuning width and decreased cortical territory for higher ordinal ranks, indicating reduced precision.
• Found that these neural responses did not generalize to symbolic ordinality, and similar tuning emerged in convolutional neural networks.

Conclusions:

• Neuronal tuning properties in parietal and premotor cortices support non-symbolic ordinality perception.
• The findings suggest that ordinality processing relies on inherent neural processing features, similar to other quantity representations.
• The emergence of similar tuning in artificial neural networks suggests a fundamental computational principle for rank processing.