Why is GABA related to neural distinctiveness? A computational account of age-related neural dedifferentiation

Neural activation patterns in response to different stimuli (e.g., houses vs. faces) are less distinctive (more similar and confusable) in older adults vs. younger adults, a phenomenon known as age-related neural dedifferentiation. A growing body of evidence suggests that the brain's major inhibitory neurotransmitter (gamma aminobutyric acid or GABA) might play a role: GABA levels decline with age throughout the brain and individual differences in GABA are significantly associated with individual differences in neural distinctiveness. But why would reduced GABA levels lead to reduced neural distinctiveness? Here, we used a biologically inspired model of koniocortex to examine how reductions in GABAergic inhibition alter neural representations. We implemented four model variants in which individual-specific GABA levels, measured with magnetic resonance spectroscopy, were used to modulate divisive normalization (alone or combined with sensory degradation), lateral inhibition (control), or sensory input strength (spurious). Neural distinctiveness was quantified from model activation patterns and compared with fMRI-derived neural distinctiveness in older adults. Model-derived specificity significantly predicted individual differences in fMRI neural distinctiveness, but only in the Divisive Normalization + Sensory Degradation Model. Model-derived specificity also mediated the relationship between empirical GABA levels and fMRI neural distinctiveness. Together, these findings provide a computationally explicit account of how age-related changes in GABA could lead to age-related neural dedifferentiation.