Fragile recurrent processing in Aphantasia: Evidence from visual pattern completion

Aphantasia - the absence of voluntary visual imagery - is typically framed as an imagery deficit, but the recurrent circuits implicated in imagery also support perception. We asked whether aphantasic individuals are more vulnerable to backward masking during visual pattern completion, a task that depends on recurrent feedback to fill in missing information from partially occluded objects. 31 self-identified aphantasic participants and 41 controls categorized briefly presented, partially occluded objects (stimulus-onset asynchronies 25-150 ms) followed by either a noise mask or a blank screen. Trial-level Bernoulli mixed-effects models estimated each group's masking cost - the within-subject unmasked-minus-masked accuracy decrement. Averaged across SOAs, the masking cost was larger in the aphantasia group (β = -0.167, z = -2.06, p = 0.039). Because recurrent processing is most constrained at the briefest target durations, we treated the 25 ms SOA as the theoretically focal condition; there the group difference was reliable (β = -0.339, z = -2.42, p = 0.015), with aphantasic participants losing roughly twice the accuracy controls did (19 vs. 10 percentage points). SOA did not, however, statistically moderate the difference (β = 0.106, p = 0.140), so the SOA gradient is consistent with, but not established by, the data. The larger masking cost reflected better unmasked performance in aphantasia that masking eliminated, not worse masked performance, and a combined model showed that the masking-cost difference was reliably larger for occluded than for intact stimuli at the focal SOA (β = -1.00, p = 0.012). These findings are consistent with recurrent processing in aphantasia being preserved but more easily disrupted by subsequent visual interference.